STRUCTURE MATTERS: HOW ORGANIZATIONAL CHARACTERISTICS AFFECT MILITARY EFFORTS

Results and Discussion

The OLS regression models in Table 2.4 include unconditional and conditional relationships between organizational characteristics and military effectiveness. Based on the results of Model 3, the sophistication of military personnel considerably influences military effectiveness in battle. Specifically, for each unit increase in the attacker’s proportion of Sophistication, attackers experience a 27 percent decrease in their proportion of battlefield casualties on average. 26 This finding suggests that militaries with

an advantage over their opponent in terms of personnel quality will operate with more efficacy in battle, which supports Hypothesis 1. This proposition is also consistent with previous research that discusses how military skill influences the number of battlefield casualties. Biddle (2007: 208) acknowledges this relationship when stating,

25 The generalized linear model uses the logit link function and binomial distribution as suggested by Baum (2008).

26 I first focus on Model 3 because it includes all control variables and excludes the interaction terms, so the coefficients of Sophistication and Stratification can be

interpreted directly.

As weapons have become more lethal, unskilled militaries’ casualty rates have grown rapidly. The net result has been a growing gap between the casualty rates of skilled militaries and those of unskilled militaries over time: technology has acted as a wedge that drives apart the real military power of the skilled and of the inept, but with much less effect on the outcomes of wars between the highly skilled.

Figure 2.1 presents the marginal effect of Sophistication and indicates that an increasing advantage in troop quality creates a substantial decrease in the proportion of attacker personnel killed in battle. In this figure, the x-axis represents the range of the Sophistication measure and the y-axis indicates the predicted proportion of battlefield losses suffered by the attacker. Looking at Figure 2.1, attackers with the largest advantage of Sophistication are predicted to experience a Loss Exchange Ratio of 40 percent, while attacking militaries with comparatively low Sophistication are expected to account for nearly 67 percent of the battle’s casualties. For example, in the battle at Tannenberg (1914), the Sophistication measure for the German attackers and Russian defender is 0.761. This indicates that Germany outspent their Russian counterparts per soldier at a rate of 3 to 1. Such an advantage in troop quality allowed Germany to conclude with less than 10 pe rcent of the battle’s casualties.

[Figure 2.1 about here]

Model 3 also indicates that the bureaucratic design affects battlefield effectiveness. On average, for each unit increase in the attacker’s proportion of Stratification the attacker experiences nearly a 75 percent reduction in LER. This finding suggests that war participants were much more efficient on the battlefield when utilizing Model 3 also indicates that the bureaucratic design affects battlefield effectiveness. On average, for each unit increase in the attacker’s proportion of Stratification the attacker experiences nearly a 75 percent reduction in LER. This finding suggests that war participants were much more efficient on the battlefield when utilizing

[Figure 2.2 about here]

Figure 2.2 presents the marginal effect of Stratification and indicates a substantial decrease in attacker’s LER as the attacker has a greater proportion of ranked positions. In this figure, the x-axis represents the range of the Stratification measure and the y-axis indicates the proportion of battlefield casualties experienced by the attacking military. Based on Figure 2.2, an attacker with the largest advantage in terms of Stratification are expected to experience an LER of roughly 11 percent, while attacking militaries with the lowest proportion of ranked positions account for more than 85 percent of the battle’s casualties. It is important to note that as Stratification approaches 0.5, belligerents can expect to bear a near equal share of casualties.

[Figure 2.3 about here]

While it is important to consider the independent influence of organizational traits on battlefield efficacy, the combination of these elements also affect how militaries perform in battle. At first glance, the conditional relationship proposed in Hypothesis 3 is While it is important to consider the independent influence of organizational traits on battlefield efficacy, the combination of these elements also affect how militaries perform in battle. At first glance, the conditional relationship proposed in Hypothesis 3 is

attacker’s share of Stratification. This figure also illustrates that a military is expected to lose the larger proportion of the battlefield when it lags behind an opponent in terms of

personnel sophistication and bureaucratic stratification.

[Figure 2.4 about here]

To evaluate this relationship in another way, I present Figure 2.4 which shows predicted probabilities of LER when belligerents have an even share of Stratification, but differ in terms of Sophistication. Based on this figure, conditions under which Sophistication and Stratification have a significant effect on the loss exchange ratio when the upper and lower bounds of the confidence interval are both above (or below) the 0.5

line (Brambor et al 2006). 27 Thus, when belligerents share similar bureaucratic stratification but the attacker has a 0.2 of Sophistication, an attacker is predicted to

experience 58 percent of a battle’s casualties. In contrast, when the attacking military accounts for 0.8 of the dyad’s Sophistication, it can expect to experience less than 42 percent of the battle’s casualties.

27 The 0.5 line represents the null hypothesis in which sophistication and stratification have no effect on battlefield casualties.

This impact of battlefield losses takes on additional meaning when considering that more than 30 million personnel wounded or killed in the First World War (Dupuy and Dupuy 1993). While the data and substantive predictions suggest that military organizations that had more sophisticated personnel and stratified bureaucracies than their opponents operated more effectively during World War I, this does not mean they were immune from experiencing a large number of casualties. Even militaries widely perceived as professionalized and effective (e.g. Germany), did not achieve fewer personnel losses than its opponent in every battle. In fact, Murray (2011) suggests that a common trend among World War I participants was that they focused less on how to minimize battle losses, but instead developed ways to help their troops tolerate casualties. In Table 2.5, I present the battle efficacy of each war participant.

[Table 2.5 about here]

In addition to organizational traits, the relative share of democracy also affects battlefield efficacy significantly. For e very unit increase in the attacker’s proportion of the democracy score, the attacker is expected to endure a 59 percent reduction in LER on average. This finding is consistent with prior research that claims that military organizations are microcosms of the societies they serve in terms of cultural norms and initiative on the battlefield (Reiter 2007; Reiter and Stam 2002). Moreover, the statistical findings support the idea that democracies produce relatively more effective military personnel (Horowitz et al. 2011; Reiter 2007; Reiter and Stam 2002).

The presence of a military coalition also improves military effectiveness. Specifically, the model indicates that on average, attackers that operate within a coalition framework experience a 21 percent reduction in the proportional loss of military The presence of a military coalition also improves military effectiveness. Specifically, the model indicates that on average, attackers that operate within a coalition framework experience a 21 percent reduction in the proportional loss of military

Conclusion

This chapter presents a novel explanation for the variation in battlefield effectiveness witnessed in the First World War by examining the organizational structures of participating militaries. Accounting for organizational characteristics is essential because organizational structure influences key factors of the military including the development of group cohesion, mobilization and training of personnel, and the institution a bureaucratic hierarchy (Millett et al. 1988; Soeters et al. 2010; Wilson 1989). The empirical tests indicate that both personnel quality and stratification of the bureaucratic hierarchy influences battlefield efficacy. Specifically, military organizations with more sophisticated personnel and a larger command chain than their opponents experienced significantly fewer battlefield casualties.

There is a danger that the present study attempts to generalize too much information from a single instance of war. It would difficult to argue that actions and outcomes of World War I have direct implications for all other wars in history. While a centralized bureaucracy may have aided efficacy in this particular time period, it may no longer be applicable to contemporary warfare involving advanced techniques and technologies (see Biddle 2007; Murray 2011; Soeters et al. 2010; Wilson 1989). Nevertheless, scholars continue to examine the First World War because it provides a case in which participants of the conflict experiences considerable variation in terms of There is a danger that the present study attempts to generalize too much information from a single instance of war. It would difficult to argue that actions and outcomes of World War I have direct implications for all other wars in history. While a centralized bureaucracy may have aided efficacy in this particular time period, it may no longer be applicable to contemporary warfare involving advanced techniques and technologies (see Biddle 2007; Murray 2011; Soeters et al. 2010; Wilson 1989). Nevertheless, scholars continue to examine the First World War because it provides a case in which participants of the conflict experiences considerable variation in terms of

The conclusions of this study may also have implications for conflicts involving coalitions and alliances, in which various military organizations must cooperate and coordinate actions (Rice 1997). While military organizations may benefit by altering their bureaucratic structure to comport with potential coalition partners, implementing such reforms present a challenge. By their nature, bureaucratic organizations favor slow, incremental changes in behavior to create predictability and stability within its ranks (Allison 1971; Wilson 1989). Moreover, differences in the sophistication of troops would limit the ability of coalition partners to develop common methods of addressing a shared concern. Benasahel (2007: 196) recognizes this challenge when noting,

The most capable military may not be able to execute operations in its preferred manner if it is operating as part of an alliance: it may have to adjust the very qualities that make it so capable to accommodate its allies and partners. It may not be possible, for example, to execute highly flexible and adaptive operations when inflexible and static partners are present on the same battlefield, or to use the full capabilities of advanced technologies alongside militaries that cannot operate in a similar manner.

Whether acting unilaterally or as part of a multilateral alignment, organizational structure influences the ability of a military to implement actions effectively on the Whether acting unilaterally or as part of a multilateral alignment, organizational structure influences the ability of a military to implement actions effectively on the

Tables and Figures

Table 2.1: WWI Participants in Data

Country

Total Austria-Hungary

1 1 2 United Kingdom

21 7 28 United States

22 2 24 Note: Belgium, Bulgaria, Greece, Japan, Portugal, and Romania are not included in the

data.

Table 2.2: Descriptive Statistics

Std. Dev.

0.167 0.833 Human Capital

0.000 1.000 Note: LER refers to loss exchange ratio

102

0.039

0.195

Table 2.3: Organizational Structures of WWI Participants Country

Per Soldier Spending (1917) Military Ranks Austria-Hungary

15 United Kingdom

107.6218

18 United States

1715.679

17 Note: I compile the number of military ranks from Bennett (2013).

1022.646

Table 2.4: Military Organizations and Battlefield Efficacy

Model 3 Model 4 Sophistication

Military Size

(0.207) (0.207) Battle Duration

14.261 12.527 Notes: Ordinary Least Squares Regression

The dependent variable is the loss exchange ratio for each battle-dyad. Interaction refers to (Sophistication x Stratification) Robust standard errors in parentheses. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Table 2.5: Battlefield Efficacy of WWI Participants Country

Effective Defender Overall Efficacy Austria-Hungary

Effective Attacker

1 0 01/02 (50%) United Kingdom

11 4 15/28 (54%) United States

2 1 03/24 (13%) Notes: I categorize states based on the loss exchange ratio of the battle.

I consider a military organization effective if it experienced 50 percent or less of a battle’s casualties. Overall Efficacy calculates the number of “effective battles” out of total battles.

Figure 2.1: Marginal Effect of Sophistication on Battlefield Efficacy (OLS Model 3)

Note: This represents change in Sophistication with all other variables held at their mean.

Figure 2.2: Marginal Effect of Stratification on Battlefield Efficacy (OLS Model 3)

Note: This represents change in Stratifcation with all other variables held at their mean.

Figure 2.3: Predictive Margins of Interaction on Battlefield Efficacy (OLS Model 4)

Note: Predicted margins are conditional on values of Sophistication and Stratification .

Figure 2.4: Marginal Effect of Interaction on Battlefield Efficacy (OLS Model 4)

Notes: The figure illustrates the predicted LER at different levels of Sophistication if Stratification is held at 0.5. The red line identifies a 0.5 share of Loss Exchange Ratio.

Chapter 3: Cooperation, Coordination, and Alliance Formation

In the aftermath of the Second World War, many European states struggled to rebuild their economic and political infrastructures while also maintaining a military capable of addressing security concerns. Limited resources coupled with fears of increased Soviet influence prompted the United States, Canada, and much of Western Europe to form the North Atlantic Treaty Organization (NATO) in April 1949. The NATO alliance established a collective defense agreement for all signatories and a framework for allied consultation regarding threats and matters of security (NATO 2014). In order to harmonize military abilities among the allies, NATO members agreed to develop common standards for training in weapons technology to ease coordination efforts necessary for conducting joint military operations (Bensahal 2007; Leeds and Anac 2005). Additionally, NATO members adopted an alliance-wide command structure that operated independently of individual states (Bensahal 2007; Leeds and Anac 2005; Wallace 2008). The continued efforts by NATO to institutionalize best practices and procedures for its allied military personnel demonstrate that coordination, not simply

cooperation, is necessary for an alliance to function effectively. 28 While states can attract potential allies initially out of common foreign policy concerns and mutual interests, a

cooperative relationship does not guarantee smooth integration of military personnel or collaborative implementation of allied operations. Because states are unlikely to enter into a costly alliance covenant with incompatible partners, how do considerations of military coordination influence patterns of alliance formation?

28 Following the Membership Action Plan, current NATO members evaluate the military organizations of aspiring members before approving additional allies (NATO 2014).

Military alliances require elements of cooperation and coordination in order to aggregate resources, conduct joint operations, and deter outside aggressors. 29 Cooperation

occurs when states share common interests and adopt policies that benefit at least one of the actors, while not making other members worse off (Gulati et al. 2012). The possibility for cooperation is an essential first step when identifying a potential alliance partner as state leaders are unlikely to pay the extensive costs created by formal alliances unless signatories share common interests and approaches to international problems. In order to translate mutual goals into action, alliance partners must also coordinate military activities. Such coordination entails a deliberate and orderly adjustment of practices and procedures to implement allied plans (Gulati et al. 2012). Allied militaries that have comparable organizational and professional cultures require less of a learning curve to reconcile their differences and collaborate effectively. Recognizing the roles of both cooperation and coordination, I investigate patterns of alliance formation from 1816-2007 and find that states with similar military organizations are significantly more likely to create security alliances. These results suggest that states evaluate the prospects of both cooperation and coordination with potential allies before forming alliances. This means that it is not enough for allies to agree on broad political and military objectives; they must also develop compatible organizational practices and military acumen as well.

This chapter begins by introducing existing research on alliance formation and describing how political, geographic, and temporal characteristics influence interstate cooperation. Next, I distinguish the concepts of cooperation and coordination and discuss

29 Prior research suggests that alliances are most likely to deter outside aggression when the partnership signals a high degree of cooperation and coordination to the rest of the

international system (Leeds and Anac 2005).

how each element contributes to the creation of alliance agreements and the implementation of joint military ventures. Third, I devise a theoretical explanation for military alliance formation in which states assess political and military traits of other states when identifying and pursuing compatible alliance partners. I then use statistical analyses to evaluate how state and military characteristics shape the likelihood that states form a military alliance. Finally, I expand on the importance of cooperation and coordination in the context of alliance agreements and consider how these elements factor into other forms of interstate collaboration.

Alliance Formation

In the presence of limited resources or an external threat, political leaders may choose to take up arms with other states in the form of a military alliance. Alliances are a desirable policy because they allow signatories to redirect resources away from defense organizations without compromising national security (Morrow 1991, 1993; Pressman

2008). 30 Allies can also develop economies of scale that facilitate effective collaboration by dividing security functions according to each member’s particular strengths (Kimball

2006; Morgan and Palmer 2003; Morrow 1993). To achieve these mutual gains, states must be willing and able to sacrifice a degree of decision-making autonomy and attempt to align allied interests (Morrow 1991, 1993; Pressman 2008).

States are capable of cooperating militarily without a formal alliance agreement, but the high costs associated with negotiating alliance terms demonstrate the intention of

30 Alliances create a mechanism for member states to acquire resources and security without extracting additional resources or commitments from the domestic population

(Feaver 1999; Huntington 1957; Morrow 1991).

signatories to honor their commitments (Kimball 2006; Leeds 2003; Wallace 2008). 31 The establishment of a formal agreement not only links each ally’s security to every other

member’s ambition, but also threatens the credibility of states that renege on alliance responsibilities (Leeds et al. 2000; Morrow 1993; Wallace 2008). Alliances

institutionalize channels through which signatories influence political and military decisions of their partners, so the relative power of signatories shapes the likelihood that the covenant will be honored (Morrow 1991; Leeds 2003; Pressman 2008). Consideration of these factors entails that states do not select alliance partners at random or unwittingly enter into military agreements with states that could disproportionately influence

decision-making processes (Morrow 1991; Pressman 2008). 32 Thus, evaluating the prospects for interstate cooperation is an essential prerequisite for identifying and

pursuing potential alliance partners.

Numerous studies identify characteristics that can make states more or less attractive as military allies. Some scholars propose that states sharing similar domestic political institutions are far more likely to create alliances with each other (Lai and Reiter 2000; Leeds 1999). Leaders who operate under different institutional settings may be incapable of credibly committing to particular actions in the future or adapting to changes in the international system (Leeds 1999). Specifically, democracies that are responsive to domestic pressures, such as public opinion and elections, are more likely to make

31 Forming alliances with military obligations require signatories to “tie their hands” and “sink costs” into the agreement, which should send a signal of a credible commitment to

the international system (Fearon 1997; Thyne 2006).

32 Although alliances can form between asymmetric powers, member states tend to benefit from separate issues, rather than one simply overpowering the other (Morrow

1991; Pressman 2008).

credible commitments, but may be slow or unable to make commitments in the first place (Leeds 1999). Likewise, autocratic states can adopt policies relatively quickly, but lack incentives to remain committed if leaders no longer perceive the agreement as beneficial (Leeds 1999). These fundamental differences can potentially hinder states from developing a common approach to a crisis or cooperating on and off the battlefield. 33

States are also more likely to pursue alliance agreements in the presence of a shared military threat. Alliances offer states the potential to aggregate capabilities and enhance the collective capabilities of the group, which may be necessary to either balance or deter an external aggressor (Glenn 2011; Leeds and Anac 2005; Weitsman 2003). A mutual threat incentivizes investment into the war effort and provides common ground necessary to achieve some level of cooperation (Weitsman 2003). In fact, Weitsman (2003: 82) argues that when stat es face “…a uniform external threat, it will be relatively easy for them to coordinate their goals and strategies to attain those goals.” This behavior was evident in both World Wars, where the threat of German ascension brought together states with similar domestic regimes, such as the United States and Canada, as well as those with disparate regime types, including the United Kingdom and Russia. The mutual threat should have been enough for the allies to operate as a military coalition, but they still chose to use formal agreements to institutionalize a cooperative relationship. 34 These

states were drawn together initially by a shared perception of threat, but fears of

33 These difficulties are common in ad hoc coalitions where partner states are typically unfamiliar with one another (Glenn 2011).

34 Despite sharing a substantial external threat, allies on both sides of the First and Second World Wars largely failed to develop a common strategic outlook or standardized

military practices (Wallace 2008).

abandonment or entrapment encouraged these states to formalize their relationship through an alliance agreement (Pressman 2008; Snyder 1984). 35

In situations where states share similar degrees of material and political power, a military alliance likely serves as a tool of capability aggregation to deter or defeat a common threat (Morrow 1991). This symmetric agreement initially limits the risk of entrapment or abandonment because both members of the alliance need each other equally to preserve their national security (Morrow 1991; Pressman 2008). Yet, if one ally increases its capabilities and is able to maintain its security unilaterally, the value of the alliance decreases along with the likelihood that alliance commitments will be honored (Morrow 1991; Pressman 2008). Conversely, states can use a strategy of issue- linkage to establish an alliance in the presence of a power asymmetry (Morrow 1991; Pressman 2008). The ally with substantial resources can enhance the security capabilities of the lesser ally in exchange for concessions in decision-making authority (Pressman

2008). 36 The ability to trade security for autonomy allows asymmetric military alliances to stay intact after changes in relative power, and provides states an opportunity to align military interests through the formation of economies of scale (Leeds 2003; Leeds and Anac 2005; Morrow 1991; Pressman 2008).

35 States can overcome differences through formal alliances because the cost of establishing the agreement creates increased costs in abrogating an agreement. This

increases the value of joint action (Leeds and Anac 2005; Leeds and Savun 2007).

36 A potential danger associated with an asymmetric relationship is that cooperation can transform into cooptation, leaving the weaker ally at the mercy of the stronger state’s

policy demands. Likewise, when given the assurance of a stronger state’s military support, a weaker ally may become emboldened, adopt aggressive policies, and entrap the more powerful ally in a conflict (Mearsheimer 2001; Pressman 2008).

A common thread running through each explanation is that states tend to form alliances with those that share similar domestic institutions, a mutual external threat, and foreign policy interests. By having some form of common ground, potential allies should

be able to cooperate in the negotiations of alliance agreements and the implementation of their commitments. Cooperation among states may be necessary to get potential allies to the negotiation table, but it does not create an immediate harmony of interests. In fact, states need a formalized agreement to provide military consultation or support because

they have diverse interests. 37 Even though alliances can foster cooperation among members, this relationship requires the joint pursuit of goals that benefit some partners

and do not make allies worse off than before the agreement was established. Intra- alliance cooperation entails that signatories share a common objective, but they may not adopt the same vision of how to pursue the goal.

Previous literature argues that cooperation is essential for allies to agree on policies in principle, but it largely overlooks how prospects of coordination factor into alliance formation decisions. The role of coordination is critical because cooperation alone does not guarantee the capability of implementing allied plans if a conflict or crisis occurs. Specifically, allied states must also coordinate their practices, procedures, and military activities in order to operate as a cohesive unit (Weitsman 2014).

Cooperation and Coordination

Scholars often use the words “cooperation” and “coordination” interchangeably, but each term denotes a distinct concept. Cooperation refers to a “joint pursuit of agreed-

37 If allies had identical interests, they would come to the aid of each other without the cost of a formal alliance agreement in order to pursue their common goals (Morrow

on goals in a manner corresponding to a shared understanding about contributions and payoffs” (Gulati et al. 2012, 533). In other words, states can cooperate by identifying a common problem and agreeing to participate in the resolution of the issue. States may be able to settle on mutual understanding of the problem at-hand, but cooperation does not guarantee a shared strategy or approach to the predicament. Even in an ideal situation where there is a perfect alignment of interests and goals, allies may have difficulty implementing joint tasks because they are unable to develop an intra-alliance division of labor or they struggle adapting cohesively to changing circumstances (Glenn 2011; Gulati et al. 2012; Weitsman 2014).

In addition to cooperation, allied states must also be able to coordinate their actions. Scholars define coordination as “the deliberate and orderly alignment or adjustment of partners’ actions to achieve jointly determined goals” (Gulati et al. 2012, 537). While cooperation entails a mutual understanding of goals, the benefit of aggregating resources, and payoffs of joint action, coordination indicates the specific ways that partners devise and implement operations (Glenn 2011; Gulati et al. 2012). Coordination focuses less on sustaining a relationship and deterring opportunistic behavior, and instead emphasizes mechanisms, such as information-sharing and standardized practices, that facilitate integration of each partner’s contributions to the alliance (Gulati et al. 2012). Put simply, coordination ensures that allied actions “click” and are able to create desired outcomes through contributions by all signatories (Gulati et al. 2012; Weitsman 2003; 2014).

By describing these concepts separately, it becomes apparent that both cooperation and coordination are essential features of collaborative efforts. Cooperation By describing these concepts separately, it becomes apparent that both cooperation and coordination are essential features of collaborative efforts. Cooperation

a mutual understanding of the task at-hand. Recognizing the extensive interplay between cooperation and coordination, I contend that alliances form when states cooperate to create a common set of interests and have military organizations capable of coordinating actions.

Choosing Alliance Partners

States expect their military organizations to provide essential security needs such as ensuring territorial integrity, promoting domestic order, and projecting state interests beyond national borders. Most militaries operate toward a similar mission, but the professional nature of their actions and approaches to issues on the battlefield reflect distinct norms and behaviors of the society that they serve (Reiter 2007; Reiter and Stam 2002). The share of resources devoted to the military not only shapes the capabilities of the armed forces, but also demonstrates how the state prioritizes the security apparatus (Brooks 2007a; Huntington 1957; Soeters et al. 2010). Because states operate with a finite amount of distributable goods, a continual competition for resources emerges between the military and other government agencies to maintain organizational relevance and vitality (Allison 1971; Barnett and Finnemore 2004).

This competitive environment places constraints on military action by defining acceptable performance metrics and making allocations of resources and responsibilities conditional on the perception of the military as a central institution of the state (Allison

1971; Burk 2001; Reiter and Stam 2002; Tellis 2000). 38 The military’s share of resources also influences the professional nature of its personnel by determining the number of combatants recruited, limiting the types of weapons and technologies available, and constraining how troops are supplied and trained (Brooks 2007a; Burk 2001; Huntington

1957; Tellis 2000). 39 Insufficient numbers of troops and a lack of sophisticated personnel force military organizations to implement suboptimal strategies, operations, and tactics,

all of which increase the risk of casualties and failure in the theater of war (Biddle and Long 2004; Reiter and Meek 1999).

Combining efforts under the auspices of an alliance requires each state to relinquish some degree of autonomy, so state leaders must carefully decide if joint operations will increase the probability of achieving mutual goals without endangering the state’s relative power (Mearsheimer 2001; Pressman 2008). Recognizing that decision-makers select alliance partners in the imperfect marketplace of the international system, they must judge other states by perceived qualities and characteristics (Crescenzi

38 States that devote substantial material and human resources to its armed forces demonstrate that the military is a vital organ of the state. On the other hand, states that

allocate limited personnel and resources to their military indicate that national security is not a top priority, and in turn constrain its ability to pursue complex or large-scale operations (Allison 1971; Bensahel 2007; Feaver 1999).

39 Personnel sophistication and professionalization translates into fundamental features of

a military organization including methods of communication through the chain of command, the development of cohesion among units, and the institutionalization of norms to abide by the bureaucratic stratification of authority (Millet et al. 1986; Soeters et al. 2010; Wilson 1989).

et al. 40 2012). All other things equal, states prefer to align themselves with states that have shared foreign policy interests, are subject to similar domestic political pressures,

and are likely to honor their agreements (Crescenzi et al. 2012; Lai and Reiter 2000; Leeds 1999, 2003). States that share these traits are more likely to have common goals and mutual interests, which provides fertile ground for a cooperative relationship (Gulati et al. 41 2012).

Beyond identifying potential allies by addressing cooperation concerns, state leaders must also account for prospects of military coordination before making a formal security pact. 42 Even if potential allies share a common vision on political and security objectives, differences in aptitude, training, and professional experience necessitate negotiation and experimentation before dissimilarities can be reconciled and effective collaboration can occur (Soeters et al. 2010; Weitsman 2003, 2014). In situations where there are vast differences between capabilities and methods of behavior, allies must invest considerable effort to accommodate their partners (Bensahel 2007; Glenn 2011; Szayna et al. 2001). Lieutenant Colonel Wayne A. Silkett recognizes the potential costs associated with intra-alliance variation when stating,

Cultural differences, subtle or substantial, may easily become debilitating if not understood and appreciated. Differences in discipline, work ethic,

40 Political leaders have difficulty identifying ideal alliance partners because of imperfect information, so they often rely on observable traits to make their decisions (Morrow

41 Cooperation entails that allied states demonstrate a substantive congruence of interests, often in terms of shared security interests on a regional and global scale (Szayna et al.

42 State leaders can minimize coordination costs and the likelihood of coordination failure by seeking partners perceived to be competent and compatible in terms of resources,

organizational processes, language, and culture (Gulati et al. 2012).

class distinctions, religious requirements, standards of living, traditions – all can cause friction, misunderstanding, and cracks in cohesion (Silkett 1993, 79). 43

The structure of allied military organizations influences how other states perceive the security partnership. 44 When military personnel come from distinct professional cultures

and organizational backgrounds, administrators and practitioners within the alliance may employ different and potentially incompatible practices and procedures (Gulati et al. 2012; Silkett 1993; Soeters et al. 2010). If military organizations share similar structures and routines, they necessitate less of a learning curve to bridge cultural and professional differences due to a comparable understanding of standard operating procedures and hierarchical controls (Glenn 2011; Gulati et al. 2012; Weitsman 2014). 45 Allied militaries

that cooperate in principle but lack common military characteristics may have difficulty understanding each other’s contributions and may fail to integrate them into a cohesive

strategy or operation (Bensahel 2007; Gulati et al. 2012). 46 Moreover, allies with substantial dissimilarities in military organizations do not have an overlap in knowledge and capabilities, leaving partner states uncertain in the abilities of one another and discouraging necessary coordination and cooperative efforts (Gulati et al. 2012).

43 Lieutenant Colonel Wayne A. Silkett served as Associate Director of Military Strategy in the Department of Corresponding Studies at the US Army War College (Silkett 1993).

44 An organization’s structure refers to internal pattern of relationships, authority, and communication demonstrated by its personnel (Fredrickson 1986).

45 Shared institutional norms can allow for coordination among allies lacking a history of collaboration by providing a basis for metrics, technical and administrative meanings,

and values related to reciprocity, information sharing, and feedback mechanisms (Gulati et al. 2012).

46 While military coordination requires allies to change behaviors to pursue a mutual goal, it does not require each partner to implement identical methods of coercion

(Morrow 1986).

Previous literature argues that states create alliance agreements based on the understanding that allies are capable of cooperating. While the prospects of cooperation are a necessary step to identify potential allies, states must also consider how partners would implement terms of the agreement. Forecasts for effective coordination are notably more important in military alliances that require consultation or preparation for joint actions. Because political leaders select allies based on perceptions of other states, known characteristics of military organizations likely factor in the decision-making process. Moreover, states pursue military alliances with the intent of enhancing national security, even if this means bearing costs to coordinate actions. To minimize this burden, alliances are most likely to form between states that have militaries practicing similar procedures and accustomed to comparable training and technology. 47 This means that all other things

equal, states are more likely to form military alliances when there are similarities in military organizations. Hypothesis 1: As military organizations of two states become more similar, the dyad is more likely to form an alliance.

Military alliances offer the possibility of enhancing national security by aggregating resources with another state, but such integration does not occur by simply signing an agreement. 48 In order to realize benefits from an alliance partnership, allies

must be able to cooperate and coordinate. Cooperation is a necessary element for

47 Militaries that have similar technological standards, organizational structures, and knowledge bases can reduce uncertainties about coordination and lessens the likelihood

of incompatibilities (Gulati et al. 2012).

48 In order to mitigate fears of abandonment and entrapment, alliances develop multilateral agreements based upon narrow and relatively explicit terms and obligations

(Leeds et al. 2000).

successful alliance partnerships, noting that allied states must agree that a joint venture is an efficient method of achieving particular goals. Beyond this, allies must coordinate actions, which may entail harmonizing standard practices and procedures between their military organizations. Because states select themselves into alliance agreements, they are rational to pursue partnerships with states that share similar interests and have comparable military capabilities.

Research Design

Political leaders pursue alliances to solidify state security, but these agreements require leaders to sacrifice some decision-making authority in order to coordinate interests and practices. Before entering into a covenant that entails a great deal of cost and commitment, state leaders evaluate potential allies and establish agreements with those perceived as the most capable and compatible. This study investigates the relative importance of state characteristics by analyzing alliances formed from 1816-2007. A

dyad-year is the unit of analysis. 49 Because political decision-makers can create or modify alliance agreements at any point in time, each dyad remains in the sample, even if

they have an active alliance. 50

I compile the number of initiated alliance agreements from

49 Some scholars may argue that the sample should only include politically relevant dyads. Politically relevant dyads exclude dyads that may lack the capability to interact

with one another (Lemke and Reed 2001). Scholars often operationalize politically relevant dyads as geographically continuous states or any pair of states that includes a major power (Lemke and Reed 2001). Because states can establish asymmetric alliances and enter into these agreements for purposes beyond resource aggregation (see Morrow 1991; Pressman 2008), I conclude that limiting the sample to politically relevant dyads unnecessarily introduces selection bias into the statistical analyses.

50 Scholars may contend that dyads should be removed from the sample after they have an alliance agreement in place, but this would not be appropriate for two reasons. First,

censoring the sample ignores the possibility that domestic issues (e.g., new leadership, regime change) or international factors (e.g., conflict) would alter state decisions to form and terminate alliance agreements (see Leeds and Savun 2007). Second, because each censoring the sample ignores the possibility that domestic issues (e.g., new leadership, regime change) or international factors (e.g., conflict) would alter state decisions to form and terminate alliance agreements (see Leeds and Savun 2007). Second, because each

[Figure 3.1 about here]

Dependent Variable The establishment of an alliance agreement takes place when state leaders agree to collaborate in a specific fashion through written treaties or public proclamations (Leeds 1999, 2003). Leeds (2003) indicates that the content of a given alliance agreement determines the specific actions and level of commitment expected of signatories. For instance, scholars consider defense pacts to entail the highest level of commitment because they require alliances partners to provide military assistance to any signatory attacked by a third party (Gibler 2009; Leeds 2003). On the other hand, previous literature suggests that military consultation (i.e. entente) agreements are the lowest level of military commitment, but they still require some degree of military cooperation and coordination during a crisis (Gibler 2009; Leeds 2003). The focus of this study is on alliances requiring military collaboration, so I create three dichotomous variables that indicate when states form (1) any type of alliance agreement, (2) a defense pact, or (3) a consultation pact. Each variable is coded as 1 if an alliance is established in a dyad-year and coded 0 otherwise.

alliance agreement requires different degrees of commitment from its signatories (see Leeds 2003), it would be an atheoretical decision to only account for the first agreement between a dyad. For example, the dyad of the United Kingdom and France would be removed from the sample after a consultation pact was enacted in 1827.

Independent Variables State leaders consider the potential compatibility between its armed forces and the militaries of other states when pursuing a security alliance. These alliances require some degree of collaboration among military organizations, so states with comparable militaries are often the most attractive alliance partners. Because leaders make decisions in an imperfect marketplace of information, they rely on known metrics to approximate characteristics of potential allies (Crescenzi et al. 2012; Gulati et al. 2012; Szayna et al. 2001). As a result, lea ders assess the organizational attributes of another state’s armed forces based on resources devoted to the defense apparatus. The investment of state resources shapes how the armed forces are trained, their access to technology, and if society perceives the military as a legitimate profession (Burk 2001; Feaver 1999; Soeters et al. 2010). Moreover, military organizations with access to considerable resources are more likely to produce professionalized personnel capable of using advanced weaponry and executing complex operations (Reiter and Stam 2002; Szayna et al. 2001; Tellis 2000).

It is intuitive to evaluate military organizations based on measures of a state’s total expenditures, personnel numbers, and access to material resources, but these data do not allow for cross-national comparisons and can lead to inaccurate conclusions. Specifically, absolute measures of military spending and the size of a state’s armed forces do not account for differences in economic strength and total population. In fact, prior research argues that the proportion of resources devoted to military preparation provides

a more appropriate measure of capability and resolve than absolute measures of personnel or expenditures (Wayman et al. 1983). Acknowledging the need to use proportional data, a more appropriate measure of capability and resolve than absolute measures of personnel or expenditures (Wayman et al. 1983). Acknowledging the need to use proportional data,

In light of these constraints, I select to evaluate military organizations based on the share of state resources devoted to the armed forces. Although there is no perfect measure to assess military organizations, the number and sophistication of military personnel can both shape the capacity of a military. Therefore, I elect to use two variables to capture characteristics of military organizations. First, states that maintain large standing militaries demonstrate that security is a high priority and that it necessitates substantial participation by the population (Feaver 1999; Huntington 1957; Tellis 2000). To capture the extent of military participation in each state, I calculate the number of military personnel divided by the total population (Wayman et al. 1983). Using the proportion of military personnel to the total population allows for cross-national comparisons, which is not possible through measures of absolute military size. Because the unit of analysis is a dyad-year, I create a variable that is the absolute difference in military personnel per capita for each dyad and name it participation difference.

This variable measures a difference between proportions, so it is limited to values between 0 and 1. The distribution of this variable is highly skewed, so I transform it using

51 Previous work also suggests that measures of gross domestic product fails to account for idiosyncrasies in economic development and includes economic activity that is

irrelevant to the capacity of the defense apparatus (Wayman et al. 1983).

the natural log. 52 Taking the natural log of numbers in this range produces a scale of all negative values. 53 For example, in 2007, the participation difference between the United

States and Canada is quite small at .003, while the difference between Niger and North Korea is relatively large .047. Although these are both positive values, the natural log of these measures is -5.722 and -3.088 respectively. Thus, the most negative values of participation difference represent the most similar dyads, while the least negative figures the most dissimilar dyads.

Political leaders can also influence the sophistication of its personnel by investing in innovative training methods and advanced technology. Specifically, prior studies argue that the rate of per soldier spending indicates the type of technology, equipment, and training programs available to military personnel (Reiter and Stam 2002). Following the practices in previous research, I divide total military expenditures by the number of military personnel measure the quality and sophistication of armed forces (Powell 2012; Reiter and Stam 2002; Szayna et al. 2001). I create the variable sophistication difference, which calculates the difference of per soldier spending for each dyad-year. The distribution of this variable is skewed, so it is transformed using the natural log. 54 There

are some cases where the difference in per soldier spending is less than one, which results in negative values after the log transformation (Azad 2007). Therefore, sophistication difference contains both negative and positive values and negative values indicating

52 See Figure 3.3A in the appendix for the distribution of the participation difference variable.

53 To calculate the natural log of a value less than 1, Euler’s constant is raised to a number that is less than the zero power: a negative number (Azad 2007).

54 See Figure 3.4A in the appendix for the distribution of the transformed sophistication difference variable.

relatively similar dyads, while large positive values indicate the most different dyads. I compile military personnel, military expenditure, and total population figures from version 4.0 of the Correlates of War National Material Capabilities dataset (Singer 1987).

I present the descriptive statistics for participation difference, sophistication difference, and all other variables used in the statistical analyses in Table 3.1 below. These measures do not identify the specific strengths and weaknesses of a given military organization, but they do approximate the extent of roles and responsibilities given to the armed forces. For example, states that have the highest measures of

sophistication include the United Kingdom, the United States, and Canada. 55 Likewise, states that demonstrate a high level of participation include France, Switzerland, and Germany. It is not surprising that a number of the aforementioned states are also partners in security alliances. Thus, it is reasonable to suspect that armed forces with similar levels of participation and sophistication demonstrate comparable capabilities and organizational characteristics, which in turn encourages alliance formation. Control Variables

Prior research indicates that alliances often occur between states with similar domestic institutions (Lai and Reiter 2000; Leeds 1999). Moreover, scholars suggest that democratic states are relatively reliable alliance partners because their leaders face heavy

55 Saudi Arabia and Kuwait are also rank near the top in terms of per soldier spending. This provides credence to Powell (2012) who argues that the impact of per soldier

spending varies across regime types. He suggests that democratic regimes that increase per soldier spending will produce increase security capabilities, while autocracies will use this investment to pay off military leaders. Although it is possible for state leaders to overspend on the military, or unwisely invest in inefficient technologies and corrupt personnel, it is a general rule that militaries with higher per soldier spending demonstrate the most sophistication and troop sophistication (Reiter and Stam 2002; Szayna et al. 2001).

costs for backing down from an international agreement that may not exist in non- democratic regimes (Leeds 1999; 2003). To evaluate the influence of regime type and similarity, I create two variables using the Polity IV dataset (Marshall et al. 2014). The first is a dichotomous variable coded as 1 if both states have democratic regimes, and 0 otherwise. States are considered democracies if they have a score of 5 or above on the Polity scale. 56 To measure the effect of regime similarity, another variable is coded as the

absolute difference of aggregate Polity scores within a dyad. Greater values on this variable indicate increasingly different regimes in a dyad.

Previous literature indicates states pursue potential allies out of mutual policy interests or because they face a common threat (Crescenzi et al. 2012). As states become increasingly similar in their foreign policy portfolios, prospects for cooperation and coordination are greater, and in turn make them attractive alliance partners. To capture foreign policy resemblance, I use Signorio and Ritter’s (1999) S-Score. This variable has

a range of -1 to +1, with the most different portfolios approaching -1 and the most similar portfolios approaching +1 (Crescenzi et al. 2012; Signorio and Ritter 1999). I create a

variable that represents the absolute difference in S-Scores for each dyad in the sample. 57 In order to reap the greatest benefits from an alliance, partner states can minimize costs

associated with interstate communication and transportation. States located in close

56 I adopt this threshold for democracy to remain consistent with recent studies of alliance formation (see Crescenzi et al. 2012; Gibler 2008; Gibler and Wohlford 2006; Lai and

Reiter 2000). While other scholars have used a higher threshold for joint democracy (see Johnson and Leeds 2011; Leeds 2003; Leeds and Savun 2007), using a more restrictive cutoff does not change the statistical or substantive influence of the variable.

57 I conducted additional tests using St rezhnev and Voeten’s (2013) UN Voting similarity scores. Voting similarity is positively correlated with the S-Score, but results using this

variable are excluded because they only cover the international system from 1946-2012.

proximity not only require fewer resources to transport goods and personnel, but are likely to have common regional interests as well. Moreover, allies require less effort to intervene militarily if they share geographic proximity (Lai and Reiter 2000; Walt 1987). The measure for geography is the capital-to-capital distance (miles) for each dyad. I compile data on distance using the EUGene data generation program (Bennett and Stam 2000).

States that are considered major powers are attractive alliance partners because they have a disproportionate amount of material resources within the international system (Lai and Reiter 2000; Morrow 1991). Likewise, major powers frequently have geopolitical interests on a global scale, so they have incentives to increase their alliance networks to ensure their spheres of interest (Lai and Reiter; Mearsheimer 2001; Waltz 1979). To account for alliances involving major power, I create a dichotomous variable coded as 1 if at least one member of the dyad is a major power and 0 otherwise.

States are motivated to align security interests with one another when facing a common external threat. Alliances not only decrease the likelihood that signatories will

be attacked, but also makes allies more likely to intervene in conflict because the third party is a shared enemy (Gibler 2008; Lai and Reiter 2000). To measure the influence of

a common threat, I construct a dichotomous variable coded as 1 if both states engaged in

a militarized dispute against the same adversary within the previous 10 years and 0 otherwise. I compile dispute data from version 4.01 of the Correlates of War Militarized Interstate Dispute dataset (Ghosn et al. 2004).

The presence of bipolarity during the Cold War facilitated states aligning into major alliance structures. During this time, the United States and Soviet Union fought for The presence of bipolarity during the Cold War facilitated states aligning into major alliance structures. During this time, the United States and Soviet Union fought for

1 if the observation occurred during the Cold War era, and 0 otherwise. 58 States exist within an anarchic and competitive system, so they make the decision

to ally under conditions of limited information (Crescenzi et al. 2012). While a state can develop a history of dependability through a direct relationship with a particular state (i.e., dyadic relationships), its dealings with the rest of the international system (i.e., extra-dyadic relationships) also contribute to its image (Crescenzi et al. 2012). Because states prefer forming alliances with states that have a reputation for upholding commitments (Crescenzi et al. 2012), a potential ally’s record of cooperation (or a lack thereof), provide essential information. Therefore, states are more likely to establish alliances with states that have a long history in the international system. I account for this temporal dimension by constructing two variables that indicate the number of years and number of years squared since a dyad has existed (Carter and Signorino 2010).

[Table 3.1 about here]

Statistical Model The statistical analyses focus on the likelihood that states establish (1) a formal alliance, (2) a defense pact, or (3) a consultation pact. Because each dependent variable is dichotomous, I employ logistic regression. Logistic regression models the log odds of the dependent variable as a linear combination of the predictor variables (Hosmer and Lemeshow 2000). The logistic regression results indicate the change in the log odds of alliance formation for a one-unit increase a given variable. I use predicted probabilities

58 I code the Cold War period as the years 1946-1991.

below to illustrate the substantive impact of participation difference and sophistication difference on the likelihood of that each type of alliance is formed. To account for dyad- specific relationships, I cluster standard errors by dyad.

[Table 3.2. about here]

Results and Discussion

Results from the logistic regression models in Table 3.2 indicate that the level of societal participation in the military affects the likelihood that states establish formal alliances. As participation difference decreases in value, the dyad is significantly more likely to create a formal alliance. This result suggests that states pursue alliances with states that have similar proportions of military personnel to the total population, which supports Hypothesis 1. Figures 3.2-3.4 present the marginal effect of participation difference and illustrate the consistent decline in (1) alliance formation, (2) defense pacts, and (3) consultation pacts as the difference grows. In these figures, the x-axis represents the range of the participation difference measure and the y-axis indicates the predicted likelihood of a given alliance agreement. Looking at Figure 3.2, dyads with comparable participation levels form alliances just under 0.24 percent of the time, while dyads with high levels of differentiation only form alliances in about 0.06 percent of cases. Of the dyads that formed alliances, 54.6 percent have a measure of participation difference

below the sample mean. 59 [Figure 3.2 about here] [Figure 3.3 about here]

59 Table 3.1A in the Appendix presents models using participation difference as the only measure of military organizations.

[Figure 3.4 about here]

A similar finding occurs when considering each dyad’s sophistication difference. As sophistication difference decreases, states are much more likely to form an alliance agreement, which supports hypothesis 1. Figures 3.5-3.7 present the marginal effect of troop sophistication and illustrate the consistent decline in the formation of (1) all alliances, (2) defense pacts, and (3) consultation pacts as the difference grows respectively. In these figures, the x-axis represents the range of the sophistication difference measure and the y-axis indicates the predicted likelihood of a given alliance agreement. The breadth of the curve represents a 95% confidence interval of predicted values at a given measure of troop sophistication. I calculate predicted probabilities by changing the values of the coalition structures, while holding all other variables at their mean values. For example, based on Figure 3.5, dyads with minimal differences in troop sophistication form alliances 0.23 percent of the time, while those with very different troop sophistication form alliances with less than 0.06 percent likelihood. The importance of this finding is evident when considering that alliance formation is a relatively rare event. 60 Specifically, 65.4 percent of the dyads that established an alliance had a value of sophistication difference 61 measuring below the sample mean.

[Figure 3.5 about here] [Figure 3.6 about here]

60 Because alliances formation is an uncommon occurrence, I also use rare events logistic regression. This method did not change the statistical or substantive results of the

statistical analyses (King and Zeng 2001). I include the rare event logit results in the appendix (see Table 3.2A).

61 Table 3.3A in the Appendix presents models using sophistication difference as the only measure of military organizations.

[Figure 3.7 about here]

To elucidate the influence of military characteristics and alliance formation, I present the example of the relationship of Soviet Union and the People’s Republic of China during

the Cold War. In February 1950, these states entered into an alliance agreement that entailed military support from signatories in any instance of aggression by Japan or one of its allies (Cheng 2004). At this time, these states demonstrated a participation difference of -4.081 and sophistication difference of 7.995, which is above the sample mean for the former and well below the sample mean for the latter. These measures indicate that the Sino-Soviet security pact included military organizations with similar societal roles and personnel sophistication. Both parties upheld this contract until April 1980 when China refused to renew the alliance agreement in light of a deteriorating political and military relationship (Cheng 2004).

The decision for China and the USSR to eliminate security ties was unexpected considering that this dyad included two major powers with similar regime types and comparable foreign policy portfolios at the height of the Cold War. Previous literature would argue that these characteristics would be more than enough to foster a cooperative relationship within the dyad. Yet, when considering the prospects of military coordination, the data indicate considerable differences between Soviet and Chinese armed forces had emerged over time. The size and sophistication of each state’s military had changed considerably over the life of the alliance, largely resulting from domestic economic reforms and shifts in foreign policy priorities (Cheng 2004). 62 By 1980, this

62 During this time, post-Mao leadership had adopted a number of reforms, including establishing an economic relationship with the United States. At the same time, the 62 During this time, post-Mao leadership had adopted a number of reforms, including establishing an economic relationship with the United States. At the same time, the

demonstrated common political institutions and mutual foreign policy interests, but such means of cooperation were not enough to overcome substantial differences in their military organizations.

Beyond characteristics of military organizations, I find that several control variables significantly decrease the likelihood of alliance formation. Dyads are less likely to form alliances as their regime types become more dissimilar. This result is consistent with prior studies that indicate that states establish agreements with states sharing similar domestic institutions (Crescenzi et al. 2012; Gibler and Wolford 2006; Leeds 1999). States are also less likely to create alliances when the dyad consists of two democracies. Interestingly, when observing defense pacts and consultation pacts exclusively, the influence of joint democracy is no longer statistically significant. This finding may not demonstrate an inability of democracies to cooperate and coordinate, but instead indicate that democracies are capable of collaboration in the absence of a formal alliance agreement (Leeds and Anac 2005). Apart from regime type, states are less likely to establish alliances as the geographic distance between them increases.

A number of factors increase the likelihood of alliance agreements as well. Dyads that share similar foreign policy interests are significantly more likely to engage in alliance formation. Likewise, states that have recently participated in a militarized dispute with a common third party are more likely to form an alliance. Mutual policy interests

Soviet Union faced challenges to its expansionist policies and prepared its personnel for conflicts with the West and China (Cheng 2004).

and a shared enemy create fertile ground for cooperative opportunities and incentivize allied intervention in the presence of a crisis (Crescenzi et al. 2012; Lai and Reiter 2000). The distribution of power influences alliance formation as well. Dyads that include at least one major power as well as those occurring during the Cold War are significantly more likely to create a formal alliance. The global interests of major powers, along with the pressures of bipolarity in the Cold War each create motives for states to increase their alliance networks (Lai and Reiter 2000; Waltz 1979).

The measures of time indicate that there is significant temporal dimension related to alliance formation. The results do not indicate support for a linear effect of time, but rather a non-linear relationship between time and alliance formation. Specifically, the time-squared variable shows that there is an inverse u-shaped relationship, where adding years to a dyad’s history increases the likelihood of alliance formation, but this effect diminishes over time. This finding suggests that states can gather information about a potential ally for a certain period, but reach a point where new information has a declining influence on the likelihood of alliance formation. I demonstrate the structure of this relationship in Figure 3.8 below.

[Figure 3.8 about here]

Conclusion

This study investigates characteristics states use to evaluate the capabilities and merits of potential alliance agreements. Based on the statistical evidence, political leaders are significantly more likely to establish alliances with states that have similar military organizations. This finding is consistent for alliance agreements in general as well as This study investigates characteristics states use to evaluate the capabilities and merits of potential alliance agreements. Based on the statistical evidence, political leaders are significantly more likely to establish alliances with states that have similar military organizations. This finding is consistent for alliance agreements in general as well as

terms. This means that leaders must simultaneously identify potential allies that have comparable political institutions and foreign policy aspirations as well as military organizations with similar practices, procedures, and capabilities. Completing such a task is difficult even in highly cooperative situations because agreements on paper do not necessarily translate into effective joint behaviors on the ground. Coordination not only requires a higher degree of commitment between allies, but also necessitates that states alter their military practices and capabilities to minimize the learning curve in joint ventures. Having comparable militaries facilitates greater coordination because allies require less of a learning curve to adjust actions in a deliberate and orderly fashion. Moreover, the results support previous literature that argues that states prefer to work alongside those that share similar domestic institutions and a mutual threat. Taken together, these findings suggest that while interstate cooperation is an essential component for identifying possible allies, scholars must also consider the prospects of military coordination when investigating alliance formation behavior. 64

Allies that have a cooperative relationship can improve coordination efforts by institutionalizing commitments and reconciling differences in military organizations. NATO exemplifies this behavior by standardizing training methods, weapons and pay

63 While neutrality and non-aggression pacts do not require direct military coordination, there is statistical evidence that differences in military organizations may also influence

the decision to form these types of alliances as well. See Table 3.2A in the Appendix.

64 Future research could address this issue by viewing alliance formation as a two-step process in which states identify potential allies based on measures of cooperation, and

then select allies from this pool using metrics of potential coordination.

grades for all members of the alliance (Bensahal 2007; Leeds and Anac 2005). While this strategy may ensure a threshold of coordination among allies, it only occurs in the presence of extensive cooperation and at a high cost to signatories. NATO has worked toward these goals for over half a century, but it still struggles to achieve technical interoperability and implement a universal standard of practices and procedures (Bensahal 2007; Leeds and Anac 2005). As the number of alliance partners has swollen to 28 states, the diversity in organizational structures and professional cultures of allied militaries has made developing acceptable strategies and alliance goals an increasingly difficult task (Leeds and Anac 2005; Weitsman 2014).

The findings of this study have meaningful policy implications because they demonstrate that the degree to which a state prioritizes and invests in its armed forces not only shapes the professional and organizational culture of the military, but also how other states perceive it as a viable alliance partner. Similarity of interests is an essential first step because states must be capable of cooperating if they intend to establish credibility

and deter outside aggression (Gibler 2008; Leeds and Anac 2005). 65 If a cooperative relationship is possible, states then evaluate whose militaries are capable of comparable

actions on and off the battlefield. In the absence of perfect information, traits such as the number and professional character of military personnel serve as coordination metrics. By altering investments in military training, technology, and professionalization techniques, states shape the ability of the armed forces to produce the public good of

65 Gibler (2008) and Leeds et al. (2000) note that the majority of alliances challenged by a third party are those perceived as weak or dysfunctional.

national security, while also changing how states perceive them as a compatible military partner.

Tables and Figures

Table 3.1: Descriptive Statistics

Min Max Participation Diff.

Mean

Std. Dev.

0.000 0.211 Participation Diff. (ln)

-16.915 -1.554 Sophistication Diff.

0.000 2416237.000 Sophistication Diff. (ln)

552442 24523.400 71374.880

-5.329 14.698 Joint Democracy

1.00 Regime Difference

0.000 20.000 Foreign Policy Similarity

5.000 12347.000 Major Power

786518 4719.460 2796.850

0.000 1.000 Mutual Threat

1.000 192.000 Cold War

786518

30.735

29.626

786518 1822.317 3856.436

1.000 36864.000

Table 3.2: Military Characteristics and Alliance Formation

Consultation Participation Difference (ln)

All Alliances

(0.018) Sophistication Difference (ln)

(0.012) Joint Democracy

(0.093) Regime Difference

(0.006) Foreign Policy Similarity

(0.001) Major Power

(0.087) Mutual Threat

(0.111) Cold War

437894 Log Likelihood

14235.597 Notes: Logistic regression.

The dependent variable is coded 1 for an alliance formed in a dyad-year and 0 otherwise. Robust standard errors clustered by dyad in parentheses. AIC and BIC assess fit and complexity of models. Smaller values indicate better model fit. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Figure 3.1: Formal Alliance Agreements, 1816-2007

Note: This figure accounts for all alliance types.

Figure 3.2: Marginal Effect of Participation Difference on Alliance Formation (All)

Notes: The effect represents change in participation difference with all other variables held at their mean. Because the values of participation difference fall between 0 and 1, the natural log of these figures results in negative values. This means that dyads with asimilar proportion of military participation are at the most negative end of the scale, while those with the most dissimilar participation rates are at the least negative end of the scale.

Figure 3.3: Marginal Effect of Participation Difference on Alliance Formation (Defense)

Notes: The effect represents change in participation difference with all other variables held at their mean. Because the values of participation difference fall between 0 and 1, the natural log of these figures results in negative values. This means that dyads with a similar proportion of military participation are at the most negative end of the scale, while those with the most dissimilar participation rates are at the least negative end of the scale.

Figure 3.4: Marginal Effect of Participation Difference on Alliance Formation (Consultation)

Notes: The effect represents change in participation difference with all other variables held at their mean. Because the values of participation difference fall between 0 and 1, the natural log of these figures results in negative values. This means that dyads with a similar proportion of military participation are at the most negative end of the scale, while those with the most dissimilar participation rates are at the least negative end of the scale.

Figure 3.5: Marginal Effect of Sophistication Difference on Alliance Formation (All)

Notes: The effect represents change in sophistication difference with all other variables held at their mean. Because some values of sophistication difference fall between 0 and 1, the natural log of these figures results in negative values. This means that dyads with the most similar rate of per solider spending are at the negative end of the scale, while those with the most dissimilar participation rates are at the positive end of the scale.

Figure 3.6: Marginal Effect of Sophistication Difference on Alliance Formation (Defense)

Figure 3.7: Marginal Effect of Sophistication Difference on Alliance Formation (Consultation)

Figure 3.8: Relationship of Time-Squared and Alliance Formation

Note: The figure demonstrates a nonlinear effect of time on the likelihood of alliance formation.

Chapter 4: UN Interventions and Peacekeeper Fatalities

As the Cold War drew to an end, the United Nations (UN) expanded the scope of its peacekeeping operations beyond “traditional” roles and began intervening in conflicts

that maintained active hostilities (Doyle and Sambanis 2006; Fortna 2004; Fortna and Howard 2008). 66 An early test for this new brand of peacekeeping arose in early 1993

with the establishment of the Second UN Operation in Somalia (UNOSOM II). Coming on the heels of a U.S.-led intervention in war-torn Somalia (UNITAF), this mission consisted of nearly 30,000 peacekeeping personnel provided by 35 member states (Clarke and Herbst 1996; O’Neill and Rees 2005). The sheer size and scope of the mission not only reflected the resolve of the international community, but also provided enough military and police personnel to engage belligerent parties aggressively.

Despite its endowment of resources and boots on the ground, the conditions of UNOSOM II deteriorated quickly. In early June 1993, militia forces commanded by Somali warlord Mohamed Farrah Hassan Aideed ambushed and brutally murdered 24 Pakistani peacekeepers (Clarke and Herbst 1996). That October, an attempt to capture Aideed led to a bloody firefight that brought about the deaths of 18 American soldiers, a Malaysian peacekeeper, and more than 300 Somali militia members and civilians

(O’Neill and Rees 2005). 67 What once appeared as an opportunity to stabilize the fragile

66 Traditional peacekeeping refers to a response “to interstate crises by stationing unarmed or lightly armed UN forces between hostile parties to monitor a truce, troop

withdrawal, or buffer zone while political negotiations went forward” (Doyle and Sambanis 2006: 12).

67 The US Army Rangers killed in the Battle of Mogadishu operated under the guise of American commanders rather than conducting a UN- sanctioned operation (O’Neill and Rees 2005). Beyond these Rangers, the United States also contributed over 3,400 troops

to UNOSOM II in 1993 (Perry and Smith 2013).

situation in Somalia instead resulted in an unfulfilled mandate and 113 United Nations peacekeepers killed in the line of duty. 68

More than 20 years removed from the Battle of Mogadishu, United Nations peacekeeping operations (PKOs) have become increasingly complex, with contemporary missions calling on peacekeeping forces to separate belligerent parties, enforce ceasefire agreements, and protect the physical security of civilians and UN personnel (Bellamy et al. 2004; Bellamy and Williams 2012; Hultman et al. 2013). United Nations interventions disrupt the balance of power within a conflict zone by interceding between belligerent parties and obstructing the policy goals of combatants (Ruggeri et al. 2012; Salverda 2013). As a result, belligerents have incentives to purposefully and violently target peacekeepers in an attempt to reshape the strategic environment and encourage the UN to withdraw from their mission (Ruggeri et al. 2012; Salverda 2013; Wood et al. 2012). In light of this phenomenon, scholars suggest that that United Nations deploy larger numbers of peacekeepers, specifically armed military and police contingents, so the operation has personnel capable of creating a buffer zone between combatants and punishing belligerents that continue to use violence (Hultman et al. 2013; Kathman 2013; Wright and Greig 2012).

While the size and resources available to peacekeeping operations can shape the legitimacy and capacity of the intervention, peacekeeping contingents must be able to coordinate efforts if they are to meet mandated objectives and protect themselves from

68 UNOSOM II experienced 82 peacekeeper fatalities in 1993, 30 fatalities in 1994, and 1 fatality in 1995 resulting from malicious acts of violence (United Nations 2014). The

events of October 1993 also spurred the withdrawal of Belgian, French, and Italian contingents in early 1994 (O’Neill and Rees 2005).

harm (Fortna 2004; Salverda 2013). Developing a resilient and unified presence in a conflict zone is a considerable challenge because peacekeeping forces are ad hoc coalitions of contingents volunteered by security organizations with different methods and capabilities. Because each contributing state prepares personnel for peacekeeping tasks according to its own standards and practices, those taking part in PKOs often adhere to diverse standard operating procedures and demonstrate dissimilar battlefield aptitude. In situations where peacekeeping partners are incapable of working together, personnel lack the ability to adapt to changing circumstances, risk being perceived as inept, and fail to restore order to the conflict zone no matter how many “blue helmets” are involved (United Nations 2008). This means that the degree of intra-coalition differences affects the ability of peacekeepers to convey credible threats, separate belligerent actors, and protect their own lives. Therefore, how do organizational differences within peacekeeping coalitions influence the likelihood and magnitude of peacekeepers killed deliberately by belligerent actors?

To evaluate peacekeeping coalitions, I focus on the organizational structures of security forces that contribute personnel to an operation. Organizational structure refers to the internal pattern of relationships, authority, and communication, so variation of these traits within a peacekeeping coalition influences the time and effort necessary to construct strategies, aggregate resources, and execute operations (Fredrickson 1986; Heidenrich 1994; Weitsman 2003, 2014). I theorize that coalition partners that function under similar organizational structures are able to coordinate efforts effectively and in turn demonstrate the aptitude necessary to deter malicious attacks by belligerent parties. Through the analysis of UN peacekeeping operations from 1990-2013, I find that To evaluate peacekeeping coalitions, I focus on the organizational structures of security forces that contribute personnel to an operation. Organizational structure refers to the internal pattern of relationships, authority, and communication, so variation of these traits within a peacekeeping coalition influences the time and effort necessary to construct strategies, aggregate resources, and execute operations (Fredrickson 1986; Heidenrich 1994; Weitsman 2003, 2014). I theorize that coalition partners that function under similar organizational structures are able to coordinate efforts effectively and in turn demonstrate the aptitude necessary to deter malicious attacks by belligerent parties. Through the analysis of UN peacekeeping operations from 1990-2013, I find that

a significantly lower rate and magnitude. This finding suggests that United Nations leadership must consider characteristics of state security organizations before constructing peacekeeping coalitions and deploying personnel into a conflict zone.

This chapter begins by framing United Nations peacekeeping operations as coalition efforts and identifying how these interventions can incentivize violence toward peacekeepers. Second, I consider the challenges facing the UN in terms of recruiting and maintaining a sizable and capable peacekeeping force. Next, I present a theoretical explanation of how the organizational differences among peacekeeping contingents affect coordination and influence the propensity of combatants to target peacekeepers deliberately. Then, I employ statistical analyses to evaluate how characteristics of peacekeeping coalitions influence the likelihood and magnitude of peacekeeper fatalities in the conflict zone. Finally, I expound on the influence of organizational structure and posit how the United Nations can reconcile organizational idiosyncrasies in ongoing and future peacekeeping operations.

Peacekeepers as Targets of Violence

Scholars conceptualize war as a bargaining process in which adversaries engage in hostilities due to information disparities and credible commitment problems (Fearon 1995; Walter 2002). In this framework, adversaries calculate the probability of winning a conflict and select their behavior based on the payoffs of reaching a settlement in the present compared to fighting for a more favorable outcome in the future (Fearon 1995; Regan 2000, 2002). Belligerent actors have incentives to retain private information about their commitment to the contested issue, so combatants make decisions in an uncertain Scholars conceptualize war as a bargaining process in which adversaries engage in hostilities due to information disparities and credible commitment problems (Fearon 1995; Walter 2002). In this framework, adversaries calculate the probability of winning a conflict and select their behavior based on the payoffs of reaching a settlement in the present compared to fighting for a more favorable outcome in the future (Fearon 1995; Regan 2000, 2002). Belligerent actors have incentives to retain private information about their commitment to the contested issue, so combatants make decisions in an uncertain

Third-party interventions alter the domestic balance of power and complicate the bargaining process by introducing a new obstacle for policy outcomes desired by belligerent parties (Balch-Lindsay et al. 2008; Kathman and Wood 2011; Wood et al.

2012). 69 Despite attempts to intervene as an impartial third party, intercession by the United Nations presents a clear threat to combatant objectives, which allows

peacekeepers to become targets of violent acts (Clarke and Herbst 1996; Salverda 2013). Scholars and policy-makers often view the concepts of neutrality and impartiality as synonymous, but the former refers to a passive indifference, while the latter indicates the participant takes an active role seeking a just outcome (Salverda 2013; United Nations 2008). This becomes more than a semantic argument when considering that the Handbook of United Nations Multidimensional Peacekeeping Operations defines impartiality as “an objective and consistent execution of the mandate, regardless of provocation or challenge…” (United Nations 2003: 56). In fact, the United Nations argues that failure to implement the mandate at all costs risks undermining the credibility and legitimacy of the entire mission (United Nations 2008). This perspective suggests that UN personnel cannot claim to be neutral because they actually serve as “referees” that penalize infractions of international norms and principles established by the United Nations (Clapham 1998; United Nations 2008, 2009).

69 Inte rventions affect the state of conflict by lessening the combatants’ capacity to police the population, disrupting their ability to funnel resources to potential supporters, and

discouraging civilian support for hostile parties (Wood et al. 2012).

Peacekeepers that intercede in hopes of facilitating a just outcome may have to side with the weaker party in order to level the playing field (Clapham 1998). Such behavior is addressed by the Handbook of United Nations Multidimensional Peacekeeping Operations, which notes that “[peacekeepers] must actively pursue the implementation of their mandate even if doing so goes against the interest of one or more of the parties” (United Nations 2003: 56). By aligning themselves with one of the belligerent groups, even on a temporary basis, peacekeepers disrupt the local balance of power and become participants in a hostile domestic bargaining process (Pouligny 2006; Ruggeri et al. 2012; Salverda 2013). This disruption provides incentives for belligerents to remove the PKO from the conflict zone, especially for the group that has the most at stake (Salverda 2013; United Nations 2008, 2009). 70 As a result, belligerent parties have

incentives to target peacekeepers with deliberate acts of violence in order to destabilize the operation, force peacekeepers to remain close to their base, or even remove the foreign presence altogether (Kathman and Wood 2011;Ruggeri et al. 2012; Salverda 2013; Wright and Greig 2012).

Peacekeepers risk death by entering an active conflict zone, but malicious acts of violence are often tactfully premeditated. For example, during UNOSOM II the ambush

70 Belligerent parties will only adopt less violent strategies if they recognize that the intervener is resolved to end the conflict, has the capacity to punish factions that shirk on

agreements, and is able to offer alternative policies to resolve incompatibilities among combatants (Kathman and Wood 2011). In this environment, a third-party intermediary serves as a guarantor of sorts and allows combatants to disclose private information regarding their capabilities, preferences, and resolve to one another (Regan and Aydin 2006, Walter 2009). By effectively separating combatants and discouraging open conflict, UN interventions can provide belligerents with an opportunity to develop a mutually acceptable solution without fear of becoming vulnerable in the post-conflict period (Fearon 1995; Regan and Aydin 2006; Walter 2002, 2009).

and murder of Pakistani peacekeepers in June 1993 took place after the Pakistanis began inspecting authorized weapon storage sites (AWSS) following a survey of the area by

American forces (Alexander 2013; O’Neill and Rees 2005). The Somali militants selected their targets based on a perception that the Pakistani forces lacked discipline and

aptitude, and because they did not want to risk a failed, bloody engagement with U.S. personnel (Alexander 2013; O’Neill and Rees 2005). 71 Based on this incident, it is not

just the sheer size of a peacekeeping operation, but also the characteristics of contributing contingents that influence whether or not peacekeepers become victims of violent acts. Peacekeeper deaths have been relatively rare events, but because even one fatality cause peacekeepers to restrict their activities or leave the operation altogether, it is important to indicate when and how often peacekeeper deaths occur. Figure 4.1 indicates the number of peacekeeper fatalities during UN PKOs from 1990-2013.

[Figure 4.1 about here]

Peacekeeping Operations as Coalitions of the Willing

Rapid growth of peacekeeping operations at the end of the Cold War spurred an abundance of research focusing on the ability of PKOs to mitigate violence and restore order in the conflict zone (Fortna and Howard 2008). This wave of literature is plagued with inconsistencies, as some studies argue that peacekeeping is incapable of preventing hostilities or establishing a durable peace (Diehl et al. 1996; Jett 2001), while others claim that peacekeeping operations are successful under certain circumstances (Doyle and Sambanis 2000; Fortna 2008; Gilligan and Stedman 2003). Recent studies attempt to

71 The UN gave the Somali militia twelve hours of notice before commencing the inspection of the AWSS. The Somalis replied that they would respond to an inspection

with acts of aggression (O’Neill and Rees 2005).

reconcile these differences by accounting for the diversity of peacekeeping operations in terms of personnel commitments and force capacity (Hultman et al. 2013; Kathman 2013). This line of research highlights how the number and type of contingents shape the perception and activities of a given operation. In terms of limiting civilian deaths, numerically larger operations have better prospects for success because they are adequately equipped to intervene between warring factions, generate an effective buffer zone, and convince belligerents that future attempts at violence will be obstructed (Hultman et al. 2013; Kathman 2013). This finding suggests that PKOs with considerable numbers of armed military and police units have the training and equipment necessary to deter and repel violence by belligerent parties (Hultman et al. 2013).

The United Nations recognizes that size and type of contingents deployed influences its ability to amass resources and project a signal of legitimacy to the international community. 72 Constructing peacekeeping operations is a difficult task for

the UN because it does not maintain its own standing security force, but instead relies upon personnel volunteered and trained by member states (Holt et al. 2009). What is more, peacekeeping operations differ from conventional conflict situations because consequences of the mission do not directly influence the survival and security of the contributing states (Glenn 2011). This environment often leads to relatively weak commitments from peacekeeper contributors alongside explicit caveats that dictate when,

72 Large PKOs also indicate a high level of UN resolve because these missions are visible to domestic and international audiences and are more difficult to withdraw due to sunk

political costs (Hultman et al. 2013).

where, and how their personnel can be employed (Glenn 2011; Saiderman and Auerswald 2012). 73

The transient nature of ties that bind peacekeeping operations together enable contributor states to terminate their participation at any point in which perceived costs exceed the benefits associated with continued membership (Glenn 2011). This means that changes in mission mandate or the fickleness of state leadership can lead to a fluctuation in personnel and contributor states involved, but the composition of peacekeeping operations also responds to the ebb and flow of hostilities in the conflict zone (Clarke and Herbst 1996; Hultman et al. 2013; Salverda 2013). For instance, because the governments contributing forces are sensitive to the risks of peacekeeper fatalities, the United Nations has implemented rules of engagement that place restrictions on peacekeepers’ use of offensive actions in an attempt to reduce their exposure to direct hostilities with belligerents (Bellamy et al. 2004; Holt et al. 2009; Saiderman and

Auerswald 2012). 74 An oft-cited example of this policy includes the need to use a gradation to the use of deadly force, even when the threat appears imminent, by requiring peacekeepers to shout verbal warning to belligerents before opening fire (Holt et al. 2009; Saiderman and Auerswald 2012). Encouraging a conservative approach in the conflict zone may be politically satisfying for contributor governments, but doing so

73 The United Nations nominally controls all elements of the peacekeeping operation, but its institutional limitations ensure each contingent has considerable leeway to act at its

own discretion and regularly communicate with their home government (Bellamy et al. 2004; United Nations 2008). More practically, if contributing states perceive UN leadership as weak or the conflict environment as deteriorating, they may select to withdraw their personnel from the operation (Doyle and Sambanis 2006).

74 Some state leaders think these conservative rules of engagement actually put peacekeepers in greater danger (Saiderman and Auerswald 2012).

limits the ability of peacekeeping commanders to adopt complex or robust operational approaches (Holt et al. 2009).

Organizational Differences in Peacekeeping Coalitions

Research in the area of organizational ecology argues that organizations develop through a life cycle that includes stages of birth, growth, maturity, revival and decline (Chen 2014). In the early stages of this cycle, organizations achieve optimal efficiency through rational, technocratic control embedded within a clearly defined command structure and strict adherence to standard operating procedures (Adler and Borys 1996; Aoki 1986; Fredrickson 1986; Glenn 2011; Soeters et al. 2010). Reliance on formalized rules, standardized routines, and hierarchical control can dampen ambition for large-scale innovations, but doing so allows organizations to use old certainties to improve their performance in the short-run and avoid risks and costs endured during the trial-and-error period most prominent in the first stage of the life cycle (Chen 2014; Horowitz 2011; Kotter 2014). Thus, relatively young organizations require close management of rank- and-file personnel, which delays communication, often causes means of communication become sluggish and assessments to be ill informed, which can result in unnecessary costs of blood and treasure (Adler and Borys 1996; Kotter 2014; Wilson 1989).

Through the accumulation of knowledge from prior experiences, mature organizations are able to cultivate an operational history, develop expertise among personnel, and formalize practices and procedures that support essential functions of the organization (Adler and Borys 1996; Chen 2014; Horowitz 2011; Kotter 2014; Soeters et al. 2010; Wilson 1989). With time and experience, organizations are able to develop best practices, delegate discretionary authority, and encourage adaptation to complex and Through the accumulation of knowledge from prior experiences, mature organizations are able to cultivate an operational history, develop expertise among personnel, and formalize practices and procedures that support essential functions of the organization (Adler and Borys 1996; Chen 2014; Horowitz 2011; Kotter 2014; Soeters et al. 2010; Wilson 1989). With time and experience, organizations are able to develop best practices, delegate discretionary authority, and encourage adaptation to complex and

(Adler and Borys 1996; Aoki 1986; Brooks 2007a; Murray 2011; Soeters et al. 2010). 75 To encourage sophistication among its personnel, organizational leaders train their members to understand the overarching purpose of strategies, operations, and tactics in order to better comprehend how the actions of individual members fits into the larger

mission of the organization (Adler and Borys 1996; Aoki 1986; Kotter 2014). 76 In order for personnel to develop this type of sophistication, the organization must be able to

operate in a stable environment and develop best practices over a relatively long period of time (Adler and Borys 1996; Kotter 2014). 77

Organizational characteristics have direct implications for peacekeeping operation, because each state’s peacekeepers will behave according to their respective

organizational practices. Specifically, the organizational structures of contributed forces shape how personnel respond to authority, disseminate information, and adapt to high- stress environments (Fredrickson 1986; Soeters et al. 2010; Wilson 1989). In a coalition framework where each contributing state brings different skills, practices, and procedures to the operation, the inability to coordinate maneuvers or establish a cohesive grand

75 Depending on organizational memory can be beneficial in the short-run, but destructive to an organization in the long term because an over-emphasis on previous experiences

can cause an organization to stop updating its knowledge base and refuse adopting innovations (Chen 2014; Horowitz 2011).

76 These qualities fit definitions of professionalism described by Huntington (1957) and Fredrickson (1986).

77 Prior studies indicate that organizations comprised of rational individuals with both adequate experience and sufficient autonomy are better equipped to adapt strategies and

structures to fit rising environmental challenges (Chen 2014; Soeters et al. 2010).

strategy can lead to ill-considered and poorly executed actions in the conflict zone (Glenn 2011; Millett et al. 1986, 1988; Murray 2011). Even if each member of the peacekeeping coalition shares a common vision on political and security objectives, differences in aptitude, training, and professional experience necessitate negotiation and experimentation before variation can be reconciled and effective collaboration can occur (Soeters et al. 2010; Weitsman 2003, 2014). In situations where there are vast differences between partner capabilities and methods of behavior, the coalition expends considerable efforts to accommodate its members rather than focusing on the mission mandate (Bensahel 2007; Glenn 2011; Szayna et al. 2001). Lieutenant Colonel Wayne A. Silkett recognizes the potential costs associated with intra-coalition differences when stating,

Cultural differences, subtle or substantial, may easily become debilitating if not understood and appreciated. Differences in discipline, work ethic, class distinctions, religious requirements, standards of living, traditions – all can cause friction, misunderstanding, and cracks in cohesion (Silkett 1993, 79). 78

As members of an ad hoc coalition, the organizational characteristics of volunteer forces shapes how personnel prepare in terms of training and discipline as well as how they work with others in an adapt-or-die scenario (Doyle and Sambanis 2006; Gordon 2001). Because each peacekeeping contingent has its own means and methods to construct strategies and operations, having similar organizational structures within a coalition

78 Lieutenant Colonel Wayne A. Silkett served as Associate Director of Military Strategy in the Department of Corresponding Studies at the US Army War College (Silkett 1993).

He makes this statement with conventional military coalitions in mind, but it has merit for peacekeeping coalitions that require military, police, and observer personnel to coordinate operations (see United Nations 2008).

provides an avenue to readily reconcile diverse abilities and reduce the learning curve necessary to operate in a cohesive manner (Bensahel 2007; Heidenrich 1994; Millet et al. 1986). 79 Peacekeeping coalitions comprised of contingents with diverse organizational

characteristics require time and effort to become acclimated with one another. In the presence of considerable organizational differences, there is a heightened likelihood for friction and misunderstandings among peacekeeping contingents, which threatens the efficacy of actions in the conflict zone (Bensahel 2007; Brooks 2007a; Silkett 1993).

The inability to reconcile such differences not only presents challenges for intra- coalition relationships, but also projects a signal of ineptitude that invites belligerent parties to attack. Coordination difficulties were evident in UNOCI, where more than 50 member state contributed to the peacekeeping coalition. 80 These contingents were highly

diverse in terms of experience, training, and organizational culture. Despite having a sizable presence of nearly 10,000 security personnel on the ground during its tenure, belligerents still targeted UNOCI forces, with 7 peacekeepers from Niger ambushed and killed in June 2012 (Watkins 2012).

On the other hand, peacekeeping coalitions that overcome organizational differences create the perception of a capable, competent, and unified group, which can discourage malicious acts by belligerent parties. The benefits of sharing similar organizational traits are evident in the United Nations Peacekeeping Force in Cyprus

79 Jeffrey Pfeffer addresses this issue in saying, “People who share experiences and attitudes are more likely to like each other because they will understand each other better,

and because liking someone who is similar is self- reinforcing as it ratifies one’s own qualities …” (Pfeffer 1985: 69).

80 UNOCI refers to the United Nations Operation in Ivory Coast that began in April 2004 and is ongoing as of the writing of this article.

(UNFICYP), where most of the peacekeeping personnel hail from stable and long- standing security organizations (O’Neill and Rees 2005). Specifically, the vast majority of UNFICYP personnel have a shared experience with NATO practices and procedures, which allows for smooth operations and limited hostilities from belligerent parties

(O’Neill and Rees 2005). 81 One type of organizational structure is not necessarily better than another, but

combining contingents with dissimilar organizational characteristics can create difficulties in terms of coordinating joint efforts (Luft 2009). Specifically, security forces from relatively young organizations may have difficulty functioning side-by-side with personnel that have an extensive operational history (Luft 2009; Weitsman 2003, 2014). In cases where organizational structures differ substantively, there may not be sufficient time for peacekeeping contingents to reconcile their differences and operate effectively. Keeping this in mind, I anticipate that belligerents are more likely to use deliberate acts of violence toward peacekeepers when peacekeeping operations are comprised of security forces with dissimilar organizational structures. Likewise, I expect the number of peacekeeper fatalities to increase when contributed forces hail from diverse

organizational cultures.

Hypothesis 1: As organizational structures become more diverse in a peacekeeping coalition, the likelihood of peacekeeper fatalities will increase. Hypothesis 2: As organizational structures become more diverse in a peacekeeping coalition, the number of peacekeeper fatalities will increase.

81 UNFICYP is an ongoing operation in Cyprus that began in 1964. Belligerent actors have not killed any peacekeepers since 1981 despite a continued and sizable UN

presence.

Peacekeeping forces face a unique challenge because the UN expects them to operate as a unified entity despite being ad hoc coalitions of contingents from a variety of organizational backgrounds. Fundamental differences in organizational characteristics influence a coalition’s prospects for success because these traits translate into the nature of communication within the command chain, the perception of group cohesion among its members, and respect for the authority of the United Nations. Coalition partners that remain at odds in terms of their organizational practices and procedures will have considerable difficulty developing group cohesion or coordinate actions (Glenn 2011; Weitsman 2014). If peacekeeping coalitions fail to reconcile these differences, the use of violence becomes a viable tool for belligerent spoilers to force peacekeepers out of the conflict zone (Ruggeri et al. 2012; Salverda 2013; United Nations 2008).

Research Design

The United Nations generally sends peacekeeping operations to dangerous and desperate locales (Fortna 2004; United Nations 2008), so it is important to identify how the characteristics of coalition contingents influence when peacekeepers become targets of malicious violence. By directly engaging hostile actors, peacekeepers upset the domestic balance of power and put themselves in the crosshairs of groups that lose the most from an international presence (Hultman et al. 2013; Salverda 2013). Because contemporary peacekeeping operations are often deployed to active, hostile conflict zones, this study analyzes all United Nations peacekeeping operations from 1990-2013. 82

The source, size, and type of personnel contributed to an operation changes throughout

82 An abundance of literature explains that the number and conditions of UN PKOs has changed dramatically since the drawdown of the Cold War (see Bellamy et al. 2004;

Fortna and Howard 2008; Kathman 2013) Fortna and Howard 2008; Kathman 2013)

operations. I compile the characteristics of each peacekeeping operation from the International Peace Institute’s Peacekeeping Database (Perry and Smith 2013). Table 4.1 lists each UN mission included in the sample along with the total number of peacekeeper fatalities during its tenure.

[Table 4.1 about here]

Dependent Variable

I create two variables that measure the degree of malicious acts experienced by UN peacekeepers. These variables account for the likelihood that a peacekeeper is killed because of deliberate violence, as well as the number of fatalities. The first variable, fatality, is dichotomous and coded as 1 if one or more peacekeepers are killed within a mission-year and coded 0 otherwise. This discrete variable indicates instances where belligerents kill peacekeepers, but does not detail the scope of malicious violence facing UN personnel. In order to identify the magnitude of peacekeeper fatalities in a conflict zone, I create a second variable, fatality frequency, which provides a count of peacekeepers killed within a given mission-year. I compile data used to assess violence toward peacekeepers from United Nations documents recording peacekeeper fatalities on an annual basis for each PKO (United Nations 2014). These data disaggregates fatalities into categories of cause including illness, accidents, and malicious violence. Events coded as malicious violence indicate that peacekeepers died after combatants

83 The original IPI database is in mission-month format. Because information regarding peacekeeper fatalities is most readily available and verifiable at an annual basis, I

collapse and convert the data to a mission-year format.

purposefully and directly targeted them. For this reason, I only include malicious fatalities in the analyses. Independent Variable

Time, experience, and institutional stability are necessary for an organization to formalize best practices, so the structure of a security apparatus is largely a product of its ability to avoid large-scale disturbances that trigger a structural overhaul (Allison 1971; Horowitz 2011). In the aftermath of major system changes, a state must reconsider and revamp the structure of its security organization to adapt to new circumstances (Horowitz 2011). I measure the organizational structure of a state’s security apparatus through the creation of the variable structure. Structure refers to the number of years that have passed since a state experienced a severe disruptive event on either the domestic or the

international front. 84 This measure is adapted from the “durable” variable in the Polity IV dataset (Marshall and Jaggers 2014) and is theoretically consistent with the calculation of

“organizational age” used by Horowitz (2011). 85 Horowitz (2011) identifies major regime change or a losing effort in an interstate war as events powerful enough to topple the

status quo structure of security organizations. 86

I modify this measure by also including losses in intrastate conflict as well as the occurre nce of a successful coup d’état as a

84 Structure is calculated from a sample of all states from 1945-2013. I code states as 0 in 1945 or their first year of independence. This coding scheme is consistent with the

“durability” variable in the Polity IV dataset.

85 Scholars commonly address the concept of organizational age within the organizational ecology literature (see Chen 2014), and it has been recently adapted to political science

research (see Asal and Rethemeyer 2008; Horowitz 2011).

86 A regime change results if there is a change of three or more on the aggregate Polity score (Marshall and Jaggers 2014).

source of shock to a security organization. 87 I derive the durability of each state’s political regime using the Polity IV dataset (Marshall and Jaggers 2014). I compile each

state’s experience with intrastate and interstate conflicts using version 4.0 of the Correlates of War Intrastate Wars and Interstate Wars datasets (Sarkees and Wayman

2010). 88 I record successful coups d’états using the 2013 version of the Center for Systemic Peace’s Coup D’état Events dataset (Marshall and Marshall 2014). 89

Peacekeeping operations are multinational coalition efforts, so the ability of peacekeepers to coordinate efforts depends on the organizational compatibility among coalition members. To address intra-coalition relationships, I create a variable, coalition structures , which calculates the dispersion of structure within each peacekeeping coalition (i.e., mission-year) using the coefficient of variation. The coefficient of variation (CV) is a measure of dispersion that represents the r atio of a variable’s standard deviation to its mean (Allison 1978). The CV is an appropriate measure of variability because it allows for comparison among observations (i.e., coalitions) that have

considerably different dispersions and means (Allison 1978). 90 Based on this measure, peacekeeping coalitions with high values of coalition structures represent diverse

87 I exclude intrastate conflicts that involve third-party interveners (i.e., internationalized wars).

88 The Correlates of War Data are used because they establish a high threshold for conflict (i.e., battle deaths), which indicates a conflict of sufficient magnitude to spur

organizational change in the military or national police.

89 I selected this dataset because of its coverage of years 1945-2013 and because it has been cross-referenced with other datasets, including Powell and Thyne (see Marshall and

Marshall 2014).

90 Because the standard deviation and mean have the same unit of measure, their ratio creates a unit free measure than allows comparison among observations (Allison 1978).

Using other measures, such as variance or standard deviation would not allow for comparisons among groups with different means and/or standard deviations (Allison 1978).

organizational characteristics among partner contingents, while low values of coalition structures indicate coalition partners that have similar organizational traits.

To provide more clarity on this measure, I use the United Nations Observer Mission in Tajikistan (UNMOT) as an example. In 1994, UNMOT involved a coalition of 5 states including Austria, Bangladesh, Denmark, Jordan, and Uruguay. 91 The value of

structure for each of these security organizations is 48, 3, 49, 5, and 9 respectively. The average value for structure for this coalition is equal to 22.8, while the standard deviation is approximately 23.563. When dividing the standard deviation by the mean, the coefficient of variation for this coalition results in a value of 1.033. The CV indicates that the UNMOT coalition experienced variability in its organizational structures at a level of 103%. This measure not only indicates the degree of structural variation in the UNMOT coalition, but also serves as a metric of structural variability that allows comparisons among all peacekeeping coalitions.

Peacekeeping coalitions with a low degree of variability in terms of organizational structure require less of a learning curve to reconcile differences among contributing states, which should allow them to perform better in the conflict zone. An example of a low variability coalition is the group of contributing states in Pakistan (UNMOGIP) in 1993. The structure measure for these coalition partners ranges from 4 (Chile) to 48 (Belgium, Denmark, Finland, Norway, and Sweden) and the coefficient of variation for this coalition is 0.519. On the other end of the spectrum, an example of a high variability coalition is the group of contributors in Sierra Leone (UNAMSIL) in

91 The United Nations established UNMOT in 1994 to monitor the ceasefire agreement between the Government of Tajikistan and the United Tajik Opposition (Jett 2001).

2000. The structures within this coalition range from 0 (Croatia, Senegal, and the Russian Federation) to 55 (Canada, Denmark, Norway, Sweden, and the United Kingdom), which results in a CV of 1.126. These examples demonstrate that coalition partners in Pakistan experienced 52% variability, while coalition partners in Sierra Leone had 113% variability in terms of organizational structure. This indicates that the peacekeeping partners in UNMOGIP had relatively similar organizational structures, while the coalition in UNAMSIL had contingents operating under highly differentiated organizational structures. Figure 4.2 illustrates the distribution of coalition structures in the sample.

[Figure 4.2 about here]

Control Variables Recent research suggests that the number and type of contingents devoted to a peacekeeping operation signal the resolve of the United Nations and define the capabilities of the mission (Hultman et al. 2013; Salverda 2013). Military personnel often enforce peace agreements, intercede between combatant parties, and in some cases, punish belligerents for continuing transgressions in the aftermath of a negotiated settlement (Kathman 2013). Police units also operate near the battlefield, but focus on providing security through monitoring and protecting civilian populations in areas where the rule of law remains absent (Kathman 2013). Furthermore, military observers participate in the operation by assessing the progress of negotiations, political reforms, and ceasefire agreements (Kathman 2013). I use data from the International Peace Institute’s Peacekeeping Database to create three variables that indicate the average number of military, police, and observer personnel in a given mission-year (Perry and Smith 2013). I divide personnel figures by 100 in order to capture the influence of 100 Control Variables Recent research suggests that the number and type of contingents devoted to a peacekeeping operation signal the resolve of the United Nations and define the capabilities of the mission (Hultman et al. 2013; Salverda 2013). Military personnel often enforce peace agreements, intercede between combatant parties, and in some cases, punish belligerents for continuing transgressions in the aftermath of a negotiated settlement (Kathman 2013). Police units also operate near the battlefield, but focus on providing security through monitoring and protecting civilian populations in areas where the rule of law remains absent (Kathman 2013). Furthermore, military observers participate in the operation by assessing the progress of negotiations, political reforms, and ceasefire agreements (Kathman 2013). I use data from the International Peace Institute’s Peacekeeping Database to create three variables that indicate the average number of military, police, and observer personnel in a given mission-year (Perry and Smith 2013). I divide personnel figures by 100 in order to capture the influence of 100

[Table 4.2 about here]

The desire to amass adequate resources and gain legitimacy within the international community encourages the United Nations to recruit contingents from as many states as possible (Glenn 2011). Despite differences in capabilities, a wide-reaching coalition can potentially reach objectives at a lower cost than if states addressed them unilaterally (Glenn 2011; Weitsman 2014). The relative influence and power of contributing states indicates the importance of the mission to the United Nations and the international community at large. Participation by permanent members of the Security Council (P5) signals the resolve of major powers and solidifies the perception that peacekeepers will have the resources and experience necessary for the mission at hand

(Voeten 2005). 92 Likewise, contributor states that share a geographic region with the mission state have an inherent interest in restoring order to avoid lapses in relationships

(e.g., trade) or the contagion effect of conflict (Beardsley 2011; Buhaung and Gleditsch 2008). To capture the source of contributing states, I use the International Peace Institute’s Peacekeeping Database to create two variables that denote the average number of (1) P5 and (2) regional states that participate in a given mission-year (Perry and Smith

2013). 93 Table 4.2 indicates that peacekeeping missions experience a variety of participation from P5 and regional members, with about 2 states in each category contributing on average.

92 P5 participation also addresses the notion that most PKOs lack adequate funding and capabilities (Bellamy et al. 2004; Berman and Sams 2000; Gordon 2001).

93 I classify states into regions based on categories used by the United Nations (2013).

Previous literature has shown that belligerent parties have greater incentive to use violent acts in the presence of an intense and divisive conflict (Hultman 2007; Wood 2010; Wood et al. 2012). To account for the conflict environment, I create two variables that indicate the (1) severity of the dispute and (2) the strength of non-state combatants. To operationalize the severity of a conflict, I use the natural log of the number of battle- related deaths during a given mission-year (see Lacina and Gleditsch 2005; Wood et al. 2012). I compile battle-related deaths from v.5-2014 of the UCDP Battle-Related Deaths Dataset (Sundberg 2008). The second variable indicates the fighting capacity of non-state combatants relative to the government. This measure is an ordinal variable capturing the strength of rebel groups based on their ability to procure arms and maintain an active fighting force. I code this variable on a scale of 0 to 3 for each mission-year according to version 3.4 of the Non State Actors in Armed Conflict Dataset (Cunningham et al. 2009). The strongest non-state actors are coded as 3, while the absence of non-state actors is

coded as 0. 94 Peacekeepers may also experience a greater level of violence during the initial stage of a peacekeeping operation. Because belligerent parties cannot perceive the intentions and resolve of peacekeepers during the early months of the intervention, combatants have incentives to target UN personnel in hopes of disrupting peacekeeping activities and forcing the United Nations to withdraw its mission (Salverda 2013; Wright and Greig 2012). What is more, belligerents are likely to challenge PKOs in the early

94 The ordinal categories include (0) no rebels, (1) low, (2) moderate, (3) high rebel capacity. This variable is highly correlated with the rebel strength variable also included

in the Non State Actors in Armed Conflict Dataset. Where there are multiple non-state actors, I code the variable as the highest capacity among the rebel groups.

stages of deployment because it takes peacekeepers time to address the lawlessness and insecurities in the conflict zone (United Nations 2008). Thus, I create a dichotomous variable coded as 1 to indicate the first year of the peacekeeping mission. 95

Statistical Models The first set of statistical analyses focus on the likelihood that peacekeepers fall victim to malicious violent attacks. Because the dependent variable is dichotomous, I employ logistic regression. Logistic regression models the log odds of the dependent variable as a linear combination of the predictor variables (Hosmer and Lemeshow 2000). The logistic regression results indicate the change in the log odds of a peacekeeper fatality for a one-unit increase in a given variable. I use predicted probabilities below to illustrate the substantive impact of coalition structure on the likelihood of peacekeeper fatalities. To account for the fact that coalitions within a given PKO are unlikely independent from one another, I cluster standard errors by peacekeeping operation.

To evaluate the number of peacekeepers killed by malicious acts, the second set of analyses utilizes negative binomial regression. A Poisson model is often used for count data, but if this model is used in the presence of over-dispersed data, the standard errors can be biased and be too small (Vuong 1989). The descriptive statistics in Table 4.2 indicate that fatality frequency has a variance that exceeds the mean, so a negative

binomial model that accounts for over-dispersion is the correct choice (Vuong 1989). 96

95 If multiple missions occur in a state during the same year, I code the first year of each mission as 1 because each mission has a unique mandate, force size, and coalition

composition.

96 Some scholars argue that that the presence of over-dispersion and excessive zeroes in the dependent variable makes a zero-inflated negative binomial (ZINB) regression more

appropriate (Vuong 1989). An underlying assumption of ZINB regression is that separate processes lead to zeroes in the data, and this does not seem applicable in the present

Once again, I cluster standard errors by peacekeeping operation, and use predicted probabilities below to illustrate the substantive impact of coalition structure on the number of peacekeeper fatalities.

Results and Discussion

Results from the logistic regression models in Table 4.3 indicate that organizational structures within a peacekeeping coalition significantly affect the

likelihood that peacekeepers are victims of violent acts. 97 As the variability of coalition structures increases, peacekeepers are significantly more likely to experience fatal

attacks, which supports hypothesis 1. Figure 4.3 presents the marginal effect of coalition structures and illustrates a steady increase in the likelihood of peacekeeper fatalities as

the variability grows. 98 The x-axis represents the range of the coalition structures measure and the y-axis indicates the predicted likelihood of peacekeeper fatalities, while

the breadth of the curve represents a 95% confidence interval of predicted values at a given measure of coalition structures. I calculate predicted probabilities by changing the values of the coalition structures, while holding all other variables at their mean values. Based on Figure 4.3, coalitions at the low end of the variability scale endure peacekeeper fatalities less than 10 percent of the time, while those with highly differentiated partner organizations experience peacekeeper deaths at nearly a 30 percent likelihood. This

study. ZINB regressions specifying the presence of an ongoing conflict in the logistic regression stage yields findings statistically and substantively similar to negative binomial regressions.

97 This finding is robust across various model specifications. Recognizing that peacekeeper fatalities are relatively uncommon, I also specified models using rare event

logistic regression in the Appendix (see King and Zeng 2001), and this did not change the statistical or substantive results.

98 Figures 4.2A and 4.3A in the Appendix provide first differences plots for variables in Model 4 and Model 8 (King et al. 2000; King et al. 2001; Tomz et al. 2003).

finding is quite important when considering that the vast majority of peacekeeping coalitions demonstrate a relatively high degree of variability in terms of coalition structures . Specifically, of the 97 mission-years that experienced peacekeeper fatalities,

76 coalitions had a coefficient of variation greater than the sample mean.

[Table 4.3 about here] [Figure 4.3 about here]

The negative binomial models in Table 4.4 also indicate that variability in organizational structures significantly increases the number of peacekeepers killed in the line of duty, which supports hypothesis 2. Figure 4.4 illustrates the marginal effect of coalition structures , with the predicted number of peacekeeper fatalities increasing as structural variability grows. Looking at Figure 4.4, coalitions with low levels of organizational variation are predicted to lose about .1 peacekeeper, while those with considerable variation are projected to have approximately 0.6 peacekeeper killed. At first glance, the influence of organizational structures appears irrelevant, with even highly diverse coalitions predicted to lose less than one peacekeeper to malicious violence. The significance of this finding is more apparent when considering belligerents rarely kill large numbers of UN personnel, regardless of the operation. In fact, of the 97 mission- years experiencing peacekeeper fatalities, 45 mission-years featured a single peacekeeper death and only 12 mission-years had double-digit fatalities. Therefore, because even a small magnitude of peacekeeper deaths may encourage states to restrict actions of its contingents or withdraw their forces from the peacekeeping operation, minimizing organizational variability is critical.

[Table 4.4 about here]

[Figure 4.4 about here] These results from these models are more comprehensible when applied to a historical case. 99 Returning to UNOSOM II, the peacekeeping coalition in this operation

featured contributions from heterogeneous organizational structures such as Bangladesh, Norway, Romania, and Tunisia, and the United States. Government officials and those in the media initially touted the merits of this mission due in part to its diversity of participants (Clarke and Herbst 1996; O’Neill and Rees 2005). Despite having a broad, “globally representative” coalition, incompatibility among peacekeeping partners led to

slow decision-making, appointment of unqualified personnel, and an ineffective chain of command (Clarke and Herbst 1996). This dysfunction carried over to the conflict zone where UN officials disproportionately leaned on the United States to provide military and logistical support and largely failed to communicate clear objectives to other participating contingents (O’Neill and Rees 2005). Such behavior demonstrated a clear lack of a united

front, which invited belligerent parties in Somalia to strike the coalition at its weakest points.

These anecdotes find support in the data, which confirm that the UNOSOM II peacekeeping coalition embodied considerable variation in terms of coalition structures. In the presence of 82 peacekeeper deaths during 1993, the measure of coalition structures

99 The relationship between coalition variability and peacekeeper fatalities also remains consistent across both time and space. For example, UNMIS endured peacekeeper

fatalities in 3 of its 7 years of operation, UNAVEM II-III experienced peacekeeper deaths in 4 of 7 years, and MONUC suffered fatalities in 8 of 10 years. UNMIS was conducted in Sudan from 2005-2011, UNAVEM operations were conducted in Angola from 1991- 1997, and MONUC operated in the Democratic Republic of Congo from 1999-2010. Each of these cases included military, police, and observer contingents. In each of these missions, almost all instances of peacekeeper fatalities occurred when the coalition variability surpassed the sample mean.

is 0.725, which indicates there were substantial organizational differences among coalition partners. The next year, coalition structures increased to 0.836 and belligerents killed another 30 peacekeepers. In 1995, coalition structures decreased to a value of 0.937, and belligerents killed an additional peacekeeper. The peacekeeping coalition in Somalia appeared unable to reconcile organizational differences among its contingents and was unable to reach mandated goals or protect the lives of its personnel.

In addition to organizational characteristics, the models also indicate that the size and type of personnel deployed to a conflict zone influence the likelihood and frequency of peacekeeper fatalities. This finding is consistent with previous literature that indicates that a large peacekeeping presence provides more targets for belligerents, and in turn, increased peacekeeper fatalities (Salverda 2013). Specifically, as the number of armed military personnel increase, both the likelihood and magnitude of peacekeeper fatalities increase. Because armed military personnel are most equipped to engage belligerent actors (Hultman et al. 2013; Kathman 2013), belligerents are most likely to view them as

a threat to the domestic balance of power and engage in violent skirmishes with military contingents as a result. On the other hand, as the number of observers increases, the frequency of peacekeeper fatalities declines. Peacekeepers that serve exclusively as observers lack the mandate, equipment, and capacity to directly engage or deter belligerent parties (Hultman et al. 2013; Kathman 2013). Belligerents are aware of these limitations and are unlikely to view them as a substantial barrier to per-intervention policy goals. Thus, belligerents lack incentive to attack purposefully UN personnel in the presence of observers who cannot intercede among combatants or affect events in the conflict zone in a meaningful way.

The statistical analyses also specify that the presence of non-state actors and well as multiple rebel groups increases the number of peacekeeper fatalities. This finding is consistent with prior research that demonstrates that strong rebel groups are more likely to escalate violence toward peacekeepers in an attempt to restrict peacekeeper activities or remove the foreign presence altogether (Salverda 2013; Wright and Greig 2012). Previous literature shows that belligerent parties have greater incentive to utilize violence in the presence of divisive and intense conflicts because these conditions make reaching a mutually acceptable settlement highly difficult (Fearon 1995; Hultman 2007; Walter 2002; Wood 2010; Wood et al. 2012). If peacekeepers stand in the way of a belligerent party from capturing resources or gaining an advantage with in the delicate domestic power struggle, UN personnel become viable targets of violence. Therefore, where multiple hostile parties have resources and grievance, belligerents are much more likely to target peacekeepers with malicious acts of violence.

Conclusion

The present study indicates that organizational differences among peacekeeping contingents influence likelihood and magnitude that belligerents purposefully target and kill United Nations peacekeepers. Specifically, as the variability of organizational structures increases within peacekeeping coalitions, both the likelihood and frequency of peacekeeper fatalities increase significantly. These results indicate that in addition to the number and type of contingents deployed, the United Nations must also consider how the combination of diverse security organizations translate into coordinated efforts in the conflict zone. A peacekeeping coalition with relatively similar organizational structures The present study indicates that organizational differences among peacekeeping contingents influence likelihood and magnitude that belligerents purposefully target and kill United Nations peacekeepers. Specifically, as the variability of organizational structures increases within peacekeeping coalitions, both the likelihood and frequency of peacekeeper fatalities increase significantly. These results indicate that in addition to the number and type of contingents deployed, the United Nations must also consider how the combination of diverse security organizations translate into coordinated efforts in the conflict zone. A peacekeeping coalition with relatively similar organizational structures

Returning to the opening example, the mission in Somalia (UNOSOM II) serves as a prime example of coalition dysfunction, where the United Nations 113 peacekeepers deliberately killed in the line of duty despite having nearly 30,000 personnel volunteered by a broad multinational coalition. Coalition forces in Somalia were unable to develop cohesion in part because of UN leadership leaning heavily on the United States for logistical support, and as a result, the U.S. refused to work closely with other peacekeeping contingents (O’Neil and Rees 2005). This asymmetric burden-sharing agreement became more costly after the United States withdrew its personnel in 1994, leaving the remaining coalition members without necessary resources or a clear command and control infrastructure (O’Neil and Rees 2005). This example demonstrates that UN leadership must not only consider the sheer size of a peacekeeping operation, but also how organizational differences among coalition members helps or hinders their ability to collaborate effectively.

The need to consider organizational structures when constructing peacekeeping coalitions presents a considerable challenge to the United Nations because it has virtually no say over who offers personnel for a particular peacekeeping mission or how states train their contingents. The best option for UN leadership is to focus on integrating contingents that come from similar organizational and professional backgrounds, rather than cobble together broad peacekeeping coalitions with diverse practices and procedures. Such efforts appear to be taking place in the ongoing mission in Iraq (UNAMI), where coalition partners display low variability in terms of coalition The need to consider organizational structures when constructing peacekeeping coalitions presents a considerable challenge to the United Nations because it has virtually no say over who offers personnel for a particular peacekeeping mission or how states train their contingents. The best option for UN leadership is to focus on integrating contingents that come from similar organizational and professional backgrounds, rather than cobble together broad peacekeeping coalitions with diverse practices and procedures. Such efforts appear to be taking place in the ongoing mission in Iraq (UNAMI), where coalition partners display low variability in terms of coalition

a step further, the UN could codify and standardize a peacekeeping training regimen for its member states. 100 Each member state would still have the final decision of whether or

not to incorporate and institutionalize such standards, but this would provide a mechanism to improve collaboration among peacekeepers, regardless of their respective

domestic circumstances or experiences with warfare. 101 In other words, United Nations officials may be able to construct effective coalitions by combining security personnel

with similar organizational cultures and professional traditions.

The organizational attributes of coalition partners may also offer insights regarding the effectiveness of conventional military alliances and coalitions. Much like peacekeeping operations, multilateral military efforts have become more common since the end of the Cold War (Morey 2015; Sillket 1993). Moreover, alliances and coalitions are often used in order to obtain legitimacy from the international community and more practically, to aggregate the resources of multiple states (Glenn 2011; Weitsman 2003, 2014). Allied states may share an interest in neutralizing a mutual threat and devote considerable resources toward the mission, but if partner states fail to reconcile organizational and professional differences, such as aligning standard operating procedures and rules of engagement, they will not be able to counter an enemy effectively (Gordon 2001; Saiderman and Auerswald 2012; Weitsman 2003, 2014).

100 This method may be a necessary first step to institutionalize predictable behaviors and codify best practices (Adler and Borys 1996).

101 There are currently attempts to enforce standards of training, capabilities, and equipment for peacekeepers, but this states that cannot endure increased procurement

costs associated with standardization of best practices are resisting (Bellamy and Williams 2012).

Allied militaries that function under similar organizational structures need less time to become acclimated with each other, and they require less of a learning curve to aggregate resources during joint operations. Accounting for organizational structures may elucidate why some coalitions fail despite having the advantage of resources and a history of collaboration, while others are able to overcome such deficiencies over time.

Tables and Figures

Table 4.1: United Nations Peacekeeping Operations, 1990-2013 Mission

Fatalities MINUCI

14 UNOSOM I

114 UNAVEM I

UNAMSIL

17 UNOSOM II

0 UNPREDEP

74 UNAVEM III

UNAVEM II

2 Notes: The sample includes all missions conducted by the United Nations Department of

UNIOSIL

0 UNTSO

Peacekeeping Operations (UNDPKO) from 1990-2013. Fatalities refer to peacekeepers killed by malicious acts of violence.

Table 4.2: Descriptive Statistics

Min Max Fatality

Variable

Mean

Std. Dev.

0.000 1.000 Fatality Frequency

0.000 82.000 Coalition Structures

0.000 8.089 P5 Contributors

0.000 5.000 Regional Contributors

0.000 13.167 Conflict Severity

0.000 9.619 Rebel Capacity

0.000 3.000 Rebel Factions

0.000 6.000 First Year

0.000 1.000 Notes: Troops, Police, and Observer variables represent their respective raw values

divided by 100. Coalition Structures has two fewer observations because the operation to Georgia (UNOMIG) in 1993 and the operation to Haiti (UNTMIH) in 1998 did not include coalitions, and therefore did not have variation in organizational structures.

Table 4.3: Coalition Structures and Likelihood of Peacekeeper Fatalities

Model 3 Model 4 Coalition Structures

(0.125) (0.125) P5 Contributors

(0.133) (0.134) Regional Contributors

(0.100) (0.100) Rebel Capacity

(0.263) Rebel Factions

(0.253) Conflict Severity

First Year -0.062 (0.451)

62 62 62 62 Log Likelihood

-172.256 -172.235 Chi2

398.896 410.940 Notes: Logistic regression.

The dependent variable is coded 1 for at least one peacekeeper fatality in a mission-year and 0 otherwise. Robust standard errors clustered by peacekeeping mission in parentheses. Smaller values of AIC and BIC indicate better model fit. Troop, Police, and Observer variables represent a change in 100 personnel respectively. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Table 4.4: Coalition Structures and the Number of Peacekeeper Fatalities

Model 7 Model 8 Coalition Structures

(0.108) (0.112) P5 Contributors

(0.154) (0.157) Regional Contributors

(0.062) (0.062) Rebel Capacity

(0.182) Rebel Factions

(0.178) Conflict Severity

0.018 (0.069) First Year

62 62 62 62 Log Likelihood

-366.838 -366.722 Chi2

794.103 805.955 Notes: Negative binomial regression.

The dependent variable is a count of peacekeeper fatalities in each mission-year. Robust standard errors clustered by peacekeeping mission in parentheses. Smaller values of AIC and BIC indicate better model fit. Alpha represents the log-transformed over-dispersion measure. Significant values of this measure indicate that the negative binomial model is more appropriate than a Poisson model. Troop, Police, and Observer variables represent a change in 100 personnel respectively. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Figure 4.1: United Nations Peacekeeper Fatalities, 1990-2013

Note: UNOSOM II experienced 82 peacekeeper fatalities in 1993.

Figure 4.2: Distribution of Coalition Structures

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Figure 4.3: Marginal Effect of Structure on Likelihood of Peacekeeper Fatalities

Notes: Effect represents change in coalition structures with all other variables held at their mean. Vertical line represents UNOSOM II coalition structures in 1993.

Figure 4.4: Marginal Effect of Structure on Number of Peacekeeper Fatalities

Notes: Effect represents change in coalition structures with all other variables held at their mean. Vertical line represents UNOSOM II coalition structures in 1993.

Chapter 5: Conclusion

This project has focused on characteristics of military organizations and how these traits influence battlefield efficacy, patterns of alliance formation, and peacekeeper security. Taken together, this study indicates that the organizational structure of a military influences its methods of mobilizing and training personnel, development of group cohesion, and the ability to coordinate actions with other states. Chapter 2 evaluates organizations at the micro-level and measures structure in terms of personnel sophistication and bureaucratic design. Chapter 3 disaggregates the concepts of cooperation and coordination, and characterizes organizations at a macro-level, based on the share of human and military resources dedicated to the armed forces. Chapter 4 examines the development and maturation of security organizations, and evaluates how well dissimilar organizations operated in a coalition framework.

The key contribution of this project is that it examines how organizational characteristics have an effect on how a military functions on and off the battlefield. This finding challenges existing literature that tends to focus on characteristics of the state, such as regime type (Bennett and Stam 1996, 1998; Reiter and Stam 2002) and material resources (Mearsheimer 2001), influence how military organizations perform on the battlefield. Rather than assuming that military organizations are interchangeable units, I argue that organizational idiosyncrasies influence the ability of a military to utilize resources at its disposal and conduct operations effectively against a broad range of adversaries (Brooks 2007; Tellis 2000). By investigating distinct forms of military efforts, I not only demonstrate that organizational structure matters, but also that these characteristics are not sensitive to a particular measure or temporal domain.

A considerable amount of literature has addressed how differences in material capabilities, political institutions, and culture can influence state performance in military efforts, but these studies largely overlook the organizational characteristics of the armed forces. 102 Organizational structure is an important factor because the internal pattern of relationships, perceptions of authority, and means of communication shape the fundamental practices and procedures of a given military. Depending on its access to resources, societal role, and political support, each organization “learns” from its operational experiences and develops its own set of best practices (Chen 2014; Horowitz 2011; Soeters et al. 2010). 103 Accounting for these idiosyncrasies provides a link between state attributes and variation of outcomes in terms of military efforts. Specifically, characteristics of military organizations can help explain why some states perform more effectively on the battlefield, how states select alliance partners, and why large numbers of peacekeepers often cannot guarantee operational success.

It is also important to recognize how organizational characteristics fit into the larger international relations literature. Following the assumption that international system is anarchic, state leaders must create some form of security organization in order to maintain domestic order and ward off foreign aggressors (Huntington 1957; Feaver 1999; Mearsheimer 2001). The realist/neorealist perspective argues that obtaining and securing tangible power is paramount, and these actions require a large and capable fighting force (Mearsheimer 2001). This school of thought assumes that rational actors,

102 Some notable exceptions include Biddle 2004 and Millett et al. 1986, 1988. 103 Organizational learning refers to the ability to gather and disseminate information, coordinate among units, and provide strong leadership (Fortna and Howard 2008; Howard 2008).

and the military organizations they create, can use the same set of resources to achieve identical goals, regardless of perceived differences on the domestic front (Mearsheimer 2001). Conceptualizing states as unitary actors deemphasizes differences among these entities and “black boxes” each state’s military organization.

The present research challenges this viewpoint by indicating that differing circumstances influences the structure of a military organization, and in turn, its ability to convert resources into military assets. Specifically, each state’s armed forces adopt a unique set of organizational practices and procedures depending on its particular set of social, political, and economic circumstances. This logic aligns more closely with neoliberalism, which acknowledges the anarchic system and importance of power, but also recognizes the influence of subnational and transnational factors to the development of the state and its institutions (Keohane 2005; Keohane and Martin 1995; Nye 1988). Moreover, it comports with claims from scholars that military organizations are microcosms of the societies they serve in terms of professional norms and initiative on the battlefield (Millet et al. 1986, 1988; Murray 2011; Reiter 2007; Reiter and Stam 2002; Soeters et al. 2010). Nevertheless, while neoliberals recognize the influence of subnational and transnational factors to the development of state institutions, they do not address explicitly differences in the armed forces or the impact of these differences.

Another debate between these theoretical perspectives deals with the possibility of military cooperation among states. From the realist/neorealist perspective, states do not align with one another unless doing so can favorably shift the distribution of power at the expense of others or because aggregation of resources is necessary to deter a mutual Another debate between these theoretical perspectives deals with the possibility of military cooperation among states. From the realist/neorealist perspective, states do not align with one another unless doing so can favorably shift the distribution of power at the expense of others or because aggregation of resources is necessary to deter a mutual

collaboration is little more than a temporary marriage of convenience (Mearsheimer 2001; Waltz 1979). On the other hand, neoliberals argue that the pursuit of o ne state’s objectives does not necessarily have to oppose the goals of another (Keohane 2005; Keohane and Martin 1995). In fact, states can elect to work with others by negotiating agreements in which each participant plays to its strengths and derives benefits from others in its areas of relative weakness (Keohane 2005; Keohane and Martin 1995). Through continued interactions, states develop a familiarity with each other’s capabilities and interests, which reduces uncertainty among state actors and provides opportunities for further cooperation in the future (Ikenberry 2000; Keohane 2005).

Both schools of thought present motives and mechanisms for military collaboration, but these arguments tend to focus on attributes of the state and continue to “black box” the armed forces. This decision glosses over the fact that military organizations are distinct entities, and considerable time and effort is required in order for

a multinational force to operate as a cohesive group (Weitsman 2003, 2014). The creation of a military coalition or alliance offers the benefit of quickly increasing security through the aggregation of allied resources, but partner organizations must relinquish a degree of autonomy to reconcile conflicting interests and methods of behavior (Morrow 1993; Weitsman 2003, 2014). Moreover, military partnerships necessitate coordination of

104 Neorealists claim that states are able to balance against potential threats domestically by increasing material and military capacity (internal balancing) or through the creation

of interstate alignments (external balancing) (Waltz 1979; Morrow 1991; Mearsheimer 2001).

actions among participating groups in order to experience substantial benefits from the alignment (Millet et al.1986, 1988; Murray 2011). Therefore, more investigation on the specific role of military organizations and the influence of organizational characteristics is necessary if research is to move beyond limitations of current theoretical arguments.

Organizational characteristics also have practical policy implications because they can influence the likelihood and magnitude of lives lost in conflict. In general, political leaders are sensitive to battlefield casualties because personnel deaths signal political ineptitude and threaten a regime’s grip on power (Bennett and Stam 1996, 1998; Dixon

1976; Horowitz et al. 2011; 1996 2002). 105 This means that leaders try to adopt policies that limit personnel losses in order to retain support of the domestic population and keep positions of authority. While previous works have argued that characteristics of the state, such as access to material resources and regime type, influence military capabilities, each chapter of this study focuses on how organizational characteristics affect different military efforts.

Chapter 2 emphasizes the role of battlefield efficacy, and argues that effective military organizations minimize their losses and impose relatively high costs on their adversary. Because most leaders have an incentive to limit personnel losses, I argue that effective militaries are those that experience fewer casualties than their opponent. To test this hypothesis, I examine personnel deaths in battles during the First World War. Based on this specific case, I find military organizations with relatively stratified bureaucracies and sophisticated personnel operate more effectively on the battlefield and experience

105 Bennett and Stam (1996) argue that democracies fight in shorter wars than other regime types because democratic leaders fear that public support will wane when the

conflict is prolonged and casualties accumulate.

fewer casualties as a result. While these results are specific to World War I, this case helps illustrate the mechanism by which organizational characteristics influence battlefield performance.

Chapter 3 expands the scope of the theory and applies it to alliance formation. In this chapter, I argue that political leaders consider elements of cooperation and coordination when choosing alliance partners. Leaders that align with incompatible organizations risk losing unnecessary lives in combat because militaries may be unable to work alongside one another. This means that political leaders recognize the influence of organizational characteristics, and create alliances strategically with states that share a similar organizational structure. I examine patterns of alliance formation from 1816- 2007, and find that states are significantly more likely to create alliances with state that have similar personnel sophistication and societal participation in the military. These findings indicate that political leaders recognize organizational characteristics and consider how combining military organizations under an alliance agreement might affect battlefield efficacy if a crisis were to occur.

Chapter 4 indicates that United Nations peacekeeping operations (PKOs) function as ad hoc coalitions, and I propose that differences among participating organizations influence how well peacekeepers can protect themselves from malicious violence. Because the UN forms PKOs from volunteer forces, an operation’s success hinges on obtaining and maintaining competent and compatible personnel in the conflict zone. Nevertheless, political leaders are sensitive to casualties, so states can still withdraw from the coalition in the face of imminent danger to peacekeeping personnel. I examine PKOs from 1990-2013 and find that peacekeeping operations composed of personnel from Chapter 4 indicates that United Nations peacekeeping operations (PKOs) function as ad hoc coalitions, and I propose that differences among participating organizations influence how well peacekeepers can protect themselves from malicious violence. Because the UN forms PKOs from volunteer forces, an operation’s success hinges on obtaining and maintaining competent and compatible personnel in the conflict zone. Nevertheless, political leaders are sensitive to casualties, so states can still withdraw from the coalition in the face of imminent danger to peacekeeping personnel. I examine PKOs from 1990-2013 and find that peacekeeping operations composed of personnel from

The inclusion of organizational characteristics opens up a number of avenues for additional research. One approach to build on the present study would be to investigate how domestic political changes alter the roles and responsibilities assigned to the military. Because the military is one of many state bureaucracies, such changes could

influence personnel sophistication as well as an organization’s capability of adapting new strategies and technologies. 106 Shifting the focus to individual militaries would allow for

more fine-grained measures of organizational characteristics and a more nuanced examination of how organizations train, mobilize, and utilize their personnel. Specifically, future studies could examine the bureaucratic hierarchy not just by the number of ranked positions, but by the number of personnel in respective positions. The distribution of personnel may not only lead to different relationships among officers and enlisted soldiers, but may also shape means of communication and discretion by personnel on the battlefield.

In a related vein, future studies could also examine how organizational traits influence authority structures within alliances and coalitions. While chapters 3 examines how political leaders consider the potential for coordination when choosing alliance

106 Horowitz (2011) begins to explore how the organizational age of a military influences its willingness to adopt new practices.

partners, this does not explain how states accomplish coordination when called on to fight together. If state militaries have less difficulty assimilating with other organizations with similar structures and practices, similar organizations may also be more willing to adopt a relatively unified command and control structure. 107 Conversely, aligned organizations that have distinct behaviors and standard operating procedures are probably less willing to sacrifice any autonomy, even to an ally. This means that organizational characteristics not only affect performance on the battlefield, but also methods of communication and decision-making within the military partnership. Thus, future research could investigate cases of multilateral warfare, and identify if and how organizational characteristics influence the means of command and control adopted by military partners.

107 Morey (2015) identifies meaningful differences in command and control structures by classifying multilateral military efforts as coalitions or wars in parallel.

Appendix A: Additional Tables and Figures – Chapter 2

Table 2.1A: Cameron & Trivedi's Decomposition of IM-test (OLS Model 4)

P-Value Heteroscedasticity

Chi-Square

Degrees of Freedom

46.60 47 0.4892 Note: P-values greater than 0.05 indicate a lack of statistical significance.

Table 2.2A: Military Organizations and Battlefield Effectiveness (GLM)

Model 7 Model 8 Sophistication

Military Size

(0.886) (0.895) Battle Duration

102 Log Likelihood

-47.985 -47.983 AIC

Notes: Generalized Linear Model Logit link function and binomial distribution family The dependent variable is the loss exchange ratio for each battle-dyad. Interaction refers to Sophistication x Stratification Robust standard errors in parentheses. Smaller values of AIC and BIC indicate better model fit. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Figure 2.1A: Distribution of Battlefield Casualties

Loss Exchange Ratio

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Figure 2.2A: Distribution of Personnel Sophistication (Ratio)

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Figure 2.3A: Distribution of Bureaucratic Stratification (Ratio)

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Figure 2.4A: Distribution of Residuals (OLS Model 4)

Figure 2.5A: Marginal Effect of Sophistication on Battlefield Efficacy (GLM Model 8)

Note: The effect represents change in Sophistication with all other variables held at their mean.

Figure 2.6A: Marginal Effect of Stratification on Battlefield Efficacy (GLM Model 8)

Note: The effect represents change in stratifcation with all other variables held at their mean.

Figure 2.7A: Predictive Margins of Interaction Term on Battlefield Efficacy (GLM

Model 8)

Predictive Margins

Sophistication Stratification=.35

Note: Predicted margins are conditional on values of both Sophistication and Stratification .

Figure 2.8A: Marginal Effect of Interaction Term on Battlefield Efficacy (GLM Model 8)

Notes: The figure illustrates the predicted LER at different levels of Sophistication if Stratification is held at 0.5. The red line identifies a 0.5 share of Loss Exchange Ratio.

Appendix B: Additional Tables and Figures – Chapter 3

Table 3.1A: Participation Difference and Alliance Formation

Consultation Participation Difference (ln)

All Alliances

(0.017) Joint Democracy

(0.086) Regime Difference

(0.005) Foreign Policy Similarity

(0.001) Major Power

(0.079) Mutual Threat

(0.119) Cold War

528434 Log Likelihood

17409.521 Notes: Logistic regression.

The dependent variable is coded 1 for an alliance formed in a dyad-year and 0 otherwise. Robust standard errors clustered by dyad in parentheses. Smaller values of AIC and BIC indicate better model fit. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Table 3.2A: Military Characteristics and Alliance Formation (Rare Events)

Consultation Participation Difference (ln)

All Alliances

(0.018) Sophistication difference (ln)

(0.012) Joint Democracy

(0.093) Regime Difference

(0.006) Foreign Policy Similarity

(0.001) Major Power

(0.087) Mutual Threat

(0.111) Cold War

437894 Notes: Rare event logistic regression.

The dependent variable is coded 1 for an alliance formed in a dyad-year and 0 otherwise. Robust standard errors clustered by dyad in parentheses. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Table 3.3A: Sophistication Difference and Alliance Formation

Consultation Sophistication difference (ln)

All Alliances

(0.012) Joint Democracy

(0.093) Regime Difference

(0.006) Foreign Policy Similarity

(0.001) Major Power

(0.088) Mutual Threat

(0.109) Cold War

437900 Log Likelihood

14269.931 Notes: Logistic regression.

Table 3.4A: Military Characteristics and Alliance Formation (Neutrality and Non-Aggression)

Non-Aggression Pacts Participation Difference (ln)

Neutrality Pacts

(0.021) Sophistication difference (ln)

(0.015) Joint Democracy

(0.098) Regime Difference

(0.006) Foreign Policy Similarity

(0.001) Major Power

(0.097) Mutual Threat

(0.143) Cold War

437894 Log Likelihood

11153.891 Notes: Logistic regression.

Figure 3.1A: Defense Pact Alliances, 1816-2007

Figure 3.2A: Consultation Pact Alliances, 1816-2007

Figure 3.3A: Distribution of Participation Difference (ln)

Participation Difference (ln)

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Figure 3.4A: Distribution of Sophistication Difference (ln)

-5 0 5 10 15

Sophistication Difference (ln)

Notes: The solid line represents a normal distribution. The dashed line represents the kernel density.

Appendix C: Additional Tables and Figures – Chapter 4

Table 4.1A: Coalition Structures and Likelihood of Peacekeeper Fatalities (Rare Event Logistic Regression)

Model 11 Model 12 Coalition Structures

(0.123) (0.122) P5 Contributors

(0.130) (0.130) Regional Contributors

(0.097) (0.097) Rebel Capacity

(0.256) Rebel Factions

(0.246) Conflict Severity

First Year -0.045 (0.439)

62 62 62 62 Notes: Rare event logistic regression.

The dependent variable is coded 1 for at least one peacekeeper fatality in a mission-year and 0 otherwise. Robust standard errors clustered by peacekeeping mission in parentheses. Troop, Police, and Observer variables represent a change in 100 personnel respectively. *p<0.10, ** p<0.05, *** p<0.01 (two-tailed)

Figure 4.1A: Coalition Structures in Peacekeeping Operations, 1990-2013

PKO Coalitions

Lowess Curve

Figure 4.2A: First Differences and the Likelihood of Peacekeeper Fatalities

Notes: First differences represent a change from one standard deviation below the mean to one standard deviation above it. Variables with a * are discrete, and FD is a change from 0 to 1.

Figure 4.3A: First Differences and the Number of Peacekeeper Fatalities

Notes: First differences represent a change from one standard deviation below the mean to one standard deviation above it. Variables with a * are discrete, and FD is a change from 0 to 1.

STRUCTURE MATTERS: HOW ORGANIZATIONAL CHARACTERISTICS AFFECT MILITARY EFFORTS

Results and Discussion

Conclusion

Tables and Figures

Chapter 3: Cooperation, Coordination, and Alliance Formation

Alliance Formation

Cooperation and Coordination

Choosing Alliance Partners

Research Design

Results and Discussion

Conclusion

Tables and Figures

Chapter 4: UN Interventions and Peacekeeper Fatalities

Peacekeepers as Targets of Violence

Peacekeeping Operations as Coalitions of the Willing

Organizational Differences in Peacekeeping Coalitions

Research Design

Results and Discussion

Conclusion

Tables and Figures

Chapter 5: Conclusion

Appendix B: Additional Tables and Figures – Chapter 3

Appendix C: Additional Tables and Figures – Chapter 4

Parts

Dokumen yang terkait

SIMILAR CHANCES TO HAVE GOOD SELF-REGULATION, COLLABORATIVE LEARNING AND ACADEMIC-ACHIEVEMENT AMONG STUDENTS WITH VARIOUS ADMISSION SELECTION: A PRELIMINARY STUDY

SUB CHRONIC TOXICITY TEST FROM ALKOHOL EXTRACT PALIASA LEAVES (Kleinhovia Hospita Linn) TO HEPARLIVER AND KIDNEY OF EXPERIMENTAL MICE

REVIEW ON THE PROBLEM OF ZINC DEFFICIENCY, PROGRAM PREVENTION AND ITS PROSPECT

COMMUNITY KNOWLEDGE ABOUT MALARIA TREATMENT, BATEALIT AND MAYONG SUBDISTRICTS, JEPARA REGENCY

HUBUNGAN UMUR, PENGETAHUAN, DAN SIKAP TERHADAP HIPERTENSI PADA WANITA HAMIL DI PUSKESMAS KRUENG BARONA JAYA KABUPATEN ACEH BESAR CORRELATION OF AGE, EDUCATION AND ATTITUDE TO HYPERTENTION IN PREGNANT WOMEN AT KRUENG BARONA JAYA HEALTH CENTER, ACEH BESAR R

PT MNC SKY VISION Tbk DAN ENTITAS ANAK AND ITS SUBSIDIARY

PT MNC SKY VISION Tbk DAN ENTITAS ANAK AND ITS SUBSIDIARY

THE IMPACT OF FINANCIAL, NON-FINANCIAL, AND CORPORATE GOVERNANCE ATTRIBUTES ON THE PRACTICE OF GLOBAL REPORTING INITIATIVE (GRI) BASED ENVIRONMENTAL DISCLOSURE

INCREASED YIELD AND NUTRITIONAL VALUE OF KUMPAI GRASS (Hymenachne amplexicaulis (Rudge) Nees.) WITH ARBUSCULAR MYCORRHIZAL FUNGI AND ORGANIC FERTILIZER IN RED-YELLOW PODZOLIC SOIL

IMPLEMENTASI HONEYPOT DENGAN RASPBERRY PI SEBAGAI ALAT BANTU PENDETEKSI KEAMANAN JARINGAN DAN PENANGKAP MALWARE HONEYPOT IMPLEMENTATION WITH PI RASPBERRY AS A TOOL FOR SECURITY NETWORK DETECTION AND MALWARE CAPTURE

Dukungan

Links

STRUCTURE MATTERS: HOW ORGANIZATIONAL CHARACTERISTICS AFFECT MILITARY EFFORTS

Results and Discussion

Conclusion

Tables and Figures

Chapter 3: Cooperation, Coordination, and Alliance Formation

Alliance Formation

Cooperation and Coordination

Choosing Alliance Partners

Research Design

Results and Discussion

Conclusion

Tables and Figures

Chapter 4: UN Interventions and Peacekeeper Fatalities

Peacekeepers as Targets of Violence

Peacekeeping Operations as Coalitions of the Willing

Organizational Differences in Peacekeeping Coalitions

Research Design

Results and Discussion

Conclusion

Tables and Figures

Chapter 5: Conclusion

Appendix B: Additional Tables and Figures – Chapter 3

Appendix C: Additional Tables and Figures – Chapter 4

Parts

Dokumen yang terkait

SIMILAR CHANCES TO HAVE GOOD SELF-REGULATION, COLLABORATIVE LEARNING AND ACADEMIC-ACHIEVEMENT AMONG STUDENTS WITH VARIOUS ADMISSION SELECTION: A PRELIMINARY STUDY

SUB CHRONIC TOXICITY TEST FROM ALKOHOL EXTRACT PALIASA LEAVES (Kleinhovia Hospita Linn) TO HEPARLIVER AND KIDNEY OF EXPERIMENTAL MICE

REVIEW ON THE PROBLEM OF ZINC DEFFICIENCY, PROGRAM PREVENTION AND ITS PROSPECT

COMMUNITY KNOWLEDGE ABOUT MALARIA TREATMENT, BATEALIT AND MAYONG SUBDISTRICTS, JEPARA REGENCY

HUBUNGAN UMUR, PENGETAHUAN, DAN SIKAP TERHADAP HIPERTENSI PADA WANITA HAMIL DI PUSKESMAS KRUENG BARONA JAYA KABUPATEN ACEH BESAR CORRELATION OF AGE, EDUCATION AND ATTITUDE TO HYPERTENTION IN PREGNANT WOMEN AT KRUENG BARONA JAYA HEALTH CENTER, ACEH BESAR R

PT MNC SKY VISION Tbk DAN ENTITAS ANAK AND ITS SUBSIDIARY

PT MNC SKY VISION Tbk DAN ENTITAS ANAK AND ITS SUBSIDIARY

THE IMPACT OF FINANCIAL, NON-FINANCIAL, AND CORPORATE GOVERNANCE ATTRIBUTES ON THE PRACTICE OF GLOBAL REPORTING INITIATIVE (GRI) BASED ENVIRONMENTAL DISCLOSURE

INCREASED YIELD AND NUTRITIONAL VALUE OF KUMPAI GRASS (Hymenachne amplexicaulis (Rudge) Nees.) WITH ARBUSCULAR MYCORRHIZAL FUNGI AND ORGANIC FERTILIZER IN RED-YELLOW PODZOLIC SOIL

IMPLEMENTASI HONEYPOT DENGAN RASPBERRY PI SEBAGAI ALAT BANTU PENDETEKSI KEAMANAN JARINGAN DAN PENANGKAP MALWARE HONEYPOT IMPLEMENTATION WITH PI RASPBERRY AS A TOOL FOR SECURITY NETWORK DETECTION AND MALWARE CAPTURE

Dokumen yang Anda mencari sudah siap untuk unduhkan